ADVANCES IN CATALYSIS AND RELATED SUBJECTS
VOLUME I
This Page Intentionally Left Blank
ADVANCES IN CATALYSIS AND R...
48 downloads
1843 Views
16MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
ADVANCES IN CATALYSIS AND RELATED SUBJECTS
VOLUME I
This Page Intentionally Left Blank
ADVANCES IN CATALYSIS AND RELATED SUBJECTS VOLUME I EDITED BY
v. I.
W. G. FRANKENBURG Lancaeter,Pa.
KOMAREWSKY
Chimgo, la.
E. K. RIDEAL London,England
EDITORIAL BOARD
P. H. EMMETT Pilleburgh, Pa.
H. S..TAYLOR Princeton, N. J .
1948 ACADEMIC PRESS INC., PUBLISHERS NEW YORK. N.Y.
Copyright, 1948, by ACADEMIC PREM INC. 125 EAST2 8 RD STRBIET NEWYORE'10, N. Y. All Rights Reserved NO PABT OF T E I B BOOK MAY BB REIPRODUCED I N ANY FOBM, BY PHOTOBTAT, MICROFILM, OR ANY OTEEB IIWANS, WITHOUT WRX'ITEN PEBMIBBION F s O M THE PUBLIBHER.
P R I N T E D I N T H E U N I T E D S T A T E 8 O F A MER IC A THE MURRAY 4RIdl'TING COMPANY WAKEFIELD, MASSACHURITTB
CONTRIBUTORS TO VOLUME I
D. D. ELEY,Nash Lecturer in Biophysical Chemistry, University of Bristol, Bristol, England
P. H. EMMETT, Mellon Institute, Pittsburgh, Pennsylvania I. FANKUCHEN, Polytechnic Institute of Brooklyn, Brooklyn, New York
R. H. GRIFFITH, The Gas Light and Coke Company, London, England V. N. IPATIEFF,Universal Oil Products Company, Riverside, Illinuis M . B. JELLINEK, Linde Air Products Company, Timawanoh, New York HERMANPINES, Research and Development Laboratories, Universal Oil Products Company, Riverside, Illinois, and Department of Chemistry, Northwestern University, Evanston, Illinois LOUISSCHMERLING, Universal Oil Products Company, Riverside, Illinois H. H. STQRCH, Chief, Synthetic Liquid Fuels Research and Development Branch, Bureau of Mines, Pittsburgh, Pennsylvania HUGH S. TAYLOR, Frick Chemieal Laboratory, Princeton University, Princeton, New Jersey 1
V
This Page Intentionally Left Blank
PREFACE The use of heterogeneous catalysts as a tool in the laboratory, in preparative chemistry, and in large scale industrial production has expanded beyond any expectations during the last three decades. Catalytic processes today dominate the production of sulfuric acid, ammonia, methanol, and many other industrial products. The cracking of mineral oils, the hydrogenation, transformation, and synthesis of hydrocarbons are almost all centered around catalytic conversions carried out with many different catalysts including some of highly specific action. Many more catalyzed reactions are being carried out in batch processes and in continuous operations, in heterogeneous and in homogeneous systems. Similarly, catalysis is being used to an ever increasing extent for the production and transformation of individual organic compounds, a develop ment which started with the catalytic hydrogenation of vegetable oils and fats and which led to the acceptance and daily use of catalytic methods in the pharmaceutical and related industries. Parallel to this development, the study of natural processes in living matter has revealed the existence and the powerful selective action of an almost infinite number of biocatalysts of highly complicated chemical structures. In spite of these amazing practical successes of catalytic methods, and of our increasing knowledge of biocatalysts, only modest progress has been made in the scientific elucidation of the working mechanism and of the basic nature of catalytic action. As a consequence, a purely empirical approach is still the only safe way to search for efficient catalysts whenever the problem arises of carrying out a desirable and thermodynamically possible chemical transformation with the help of specific catalysts. The main reason for this situation seems $0 be that a full understanding of catalytic action would require, for any given case, a much deeper knowledge of the nature and action of atomic and molecular forces than we possess today. In addition, in the field of heterogeneous catalysis the fine structure of solid surfaces plays a decisive role, and much more would have to be known about the qualitative and quantitative nature of solid surfaces than we know at present. In other words, a science of catalysis has to be erected on foundations which still have to be laid, Nevertheless, the better understanding and the mastery of catalytic action is an urgent problem and every possible effort has to be made to achieve some progress toward this aim. An essential and indispensable condition for everyone who embarks on the study of catalytic reactions, '
vij
Viii
PREFACE
either for purely practical or for scientific purposes, is to become thoroughly familiar with the practical know how and the acientific results which have been accumulated in this field. This is no easy task. Most of the publications dealing with the theory as well as the practical aspects of catalytic processes are widely scattered throughout various journals and handbooks, covering the range from theoretical physics to descriptions of industrial plants. Considering this wide dispersion of papers dealing with essentially the same basic subject and considering further the practically complete separation of biocatalytic research and publications from the field of “man made” catalysts, it appeared to the editors a worthwhile undertaking to improve the possibilities and to facilitate the study of the catalytic literature by issuing, in annual sequence, collections of papers under the name Advances in Catalysis and Related Subjects. Our aim for Advances in Catalysis is to obtain contributions from scientific and industrial workers that represent complete and detailed surveys of those specific sectors in which these authors are mainly interested and in which they have worked successfully. This not only involves treatises on the specific materials and products for which catalytic reactions have been developed, but also general reports on new scientific theories and methods which promise to become valuable for a better understanding of catalytic phenomena. We shall try to neglect none of the various types of catalytic actions, such as the catalyses in the homogeneous gas phase and in liquids, although heterogeneous or surface catalysis will necessarily be a predominant feature of the Advances in view of its particular problematic aspects and its preponderant practical importance. As to biocatalysis, no publications dealing with specialized topics are being planned for the Advances in Catalysis because this field is already covered by other annual publications. However, we intend from time to time to bring reports in which the relationship and parallelism between this special field and “normal” catalysis are discussed. In view of the complicated nature and enormous variabiiity of catalytic phenomena, it will be unavoidable that conflicting and controversial interpretations will emerge in the text of the various papers. This will illustrate the unsolved difEiculties and the uncertainties which still prevail in the study of many catalytic reactions and in the discovery and production of ,catalysts, and may stimulate new idem better than an enforced and unwarranted consensus of opinion. Considering the difficult task of compiling the important features of any given sector of research in a clear but still critical fashion, the editors are greatly indebted to the response of the authors who will present the
ix
PREFACE
'developments in fields with which they are intimately familiar. It is our aim to obtain contributions from all parts of the world aa far aa the present world situation permits. After several years Advances in Catalysis ought to present a compilation of papers, written by competent specialists, which should enable anybody to obtain a fairly complete concept of the practical achievements of current and new methods and of the various theoretical approaches to the manifold problems. We hope that the Advanees will serve a useful purpose and wilI have the wholehearted support of the readers, particularly in the form of suggestions of special authors, selection of specific topics, and of other improvements. The support and encouragement given by the publishers in this undertaking is gratefully acknowledged.
W.G. FRANKENBURG Research Laboratory, General Cigar Co., Lancaster, Pa.
V. I. KOMAREWSKY Illinois Institute of Technology, Chicago, Ill.
E.K.RIDEAL
Davy Faraday 'ResearchLaboratory, The Royal Institution, London, England
October 14, 1948
This Page Intentionally Left Blank
CONTENTS TO VOLUME I . . . CONTRIBUTORS EDITORS’ PEEFACE . . . . . . . .
. . . . . . . .
. .
. . . .
. . . . . . . .
....... . . . . . . .
.
V
,
vii
The Heterogeneity of Catalyst Surfaces for Chemieorption
BY HUGHS. TAYLOR, Frick Chemical Laboratory, Princetm University, Princeton, New Jersey
I. Introduction . . . . . . . . . . . 11. Heterogeneity on Oxide Surfaces . . 111. Heterogeneity on Metal Surfaces . . IV. Evidence from the Earlier Literature V. General Observations . . . . . . . References . . . . . . , . . . .
.
. . . . . . . . . . . . . . . . . . . . . . . . , , . . . . . . . . .
. . . . . .
. . . . . . . . . . . . . . . . , . , . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. .
-4
3 9 19 25 26
Alkylation of Isoparatlhs BY V. N. IPATIEFF A N D LOUIBSCHMERLING, Universal Oil Products Company, Riverside, Illinois
I. Introduction . . . . . . . . . . . . . . . . . . . . 11. The Mechanism of the Catalytic Alkylation of Paraffins 111. Side Reactions . . . . . . , . . . . , . . . . . , . IV. Catalysts . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . , . . . . . , .
.
. . . . . . . . . . . . . . . . . . . . . . . . . , , . . , . . . .
27
32
39 43
63
Surface Area Measurements A New Tool for Studying Contact Catalysts BY P. H. EMMETT, Mellon Institute, Pittsburgh, Pennsylvania
I. Introduction . . . . . . . , . . . . . . . . . . . , . . . . . . . 11. Measurement of Surface Areas by Low Temperature Adsorption Isotherms 111. Limitations and General Critique of the Gas Adsorption Methods . . , IV. Application of Surface Area Methods in the Study of Solid Catalysts . . References . . . . . . , . . , . . . . . . . . . . . . . . . . .
65 66
77 81 89
The Geometrical Factor in Catalysis BY R. H. GRIFFITH, The Gas Light and Coke Company, London, Enghnd
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . 11. Reactions of Unsaturated Organic Molecules . . . . . . . . 111. Reactions of Acetylene . . . . . . . . . . . . . . . . . . IV. The Synthesis of Hydrocarbone from Carbon Mono~de,.and Reactions . .. . . . . . . . . . . . . . . . . . . . , . V. The Formation of Aromatic from Aliphatic Hydrocarbons . . . xi
. . . . . . . , . . . .
91 92 95
Related
. . . . . . . .
97 101
CONTENTS
Kii
VI . The Destructive Hydrogenation of Hydrocarbons . . . . . VII . The Reactions of Heterocyclic Compounds . . . . . . . . . VIII The Oxidation of Carbon Monoxide . . . . . . . . . I X . Future Development . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . .
.
. . . . . . . .
106 107 110 111 113
. . . . . . . . .
. . . .
The Fischer-Tropsch and Related Processes for Synthesh of Hydrocarbons by Hydrogenation of Carbon Monoxide
BY H. H . STORCH. Chief.
Sythelic Liquid Fwb Research and Development Branch. Bureau of Mines. Pittsburgh. Pennaylvania
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 Development of the Ruhrchemie Process . . . . . . . . . . . . . . I11. Development of Fischer-Tropsch Process Other Than the Ruhrchemie Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Processes Related to the Fiacher-Tropsch Synthesis . . . . . . . . . . V. Carbide Formation and Carbon Deposition on Iron Catalysts . . . . . VI Kinetics and Reaction Mechaniem in the Fischer-Tropsch Synthesis . . Referencea . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
-..116 136
.
123 133 138 140 153
The Catalytic Activation of Hydrogen BY D . D. ELEY.Nash Lecturer in Biophysical Chemistry. University of B+istol. Btistol. Enghnd I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Experimental Studies of Chemisorbed Hydrogen on Metals . . . . . . 111. Theoretical Studies of Chemisorbed Hydrogen on Metala . . . . . . . IV. Production of Hydrogen At.oms by Hot Tungsten . . . . . . . . . . V . Recombination of Hydrogen Atoms . . . . . . . . . . . . . . . . VI . The Parahydrogen Conversion . . . . . . . . . . . . . . . . . . . VII . The Exchange of Atoms between Deuterium Gaa and Hydridea . . . . VIII . Exchange and Hydrogenation of Unsaturated Hydrocarbons . . . . . . I X . The Fiacher-Tropsch Synthesis . . . . . . . . . . . . . . . . . . . X . Dehydrogenation . . . . . . . . . . . . . . . . . . . . . . . . XI . Cyclisntion of n-Paraffins . . . . . . . . . . . . . . . . . . . . . X I I . The Hydrogen-Oxygen Reaction . . . . . . . . . . . . . . . . . . XI11. Biological Activation of Hydrogen . . . . . . . . . . . . . . . . . XIV. General Conclusions . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . .
167 . ..
4
158 166 172 173 174 181 184 187 188 189 190 191 193 195
Isome?ization of Alkanes
BY HERMAN PINES.Research and Developmen4 Labordorim. Univeraal Oil Prodzlcts Company. Riverside. IUinoie. and Department of Chemistry. Northwestern Univeraity. Evanata. Illinois I. Introduction
........................... ..................... III . Isomerization of Pentanee . . . . . . . . . . . . . . . . . . . . . 11. Isomerization of Butanee
201
20T 224
?wi
CONTENTS
IV. Isomerization of Hexsnne . . . . . . . . . . . . . , . . . . . . . V. Isomerization of Heptanea and Higher Alkanes . . . . . . . . . References . . . . . . . . . . . .. . . . . . . . . . . . . . . ..
244 252 255
The Application of X-Ray Diffraction to the Study of Solid Catalysts
BY M. H. JELLINEK,L i d Air Products Company, Tunawanda, New York, AND I. FANKUCHEN, Polytechnic Institute of Brookly, Brookly, New York I. 11. 111. IV.
Part I. Description of Methods Introduction. . . . . . . . . . . . . . . . . . The Purpose of X-Ray Work . . . . . . . . Techniques . . . . . . . . . . . . . . . . . . Lattice Constants . . . . . . . . . . . . . . .
Part 11. I. Identification . . . . . . 11. Crystallites and Particlw . 111. Conclusion . . . . . . .
.
.
. .
.
.
257 258 260 272
Examples of Applications . . . . . . ,
References
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . '. . . . . . .
. . .
273 280 288
. . . .
288
Part I11
. . . .,. . .
AUTHOR INDEX . . . . . . . . . . . . . SUBJECTINDEX . . . . . , , . . . . . .
. .
. . . . . .
. . ,
. .
. . . . .
. . . .
.
,
29 1 298
This Page Intentionally Left Blank
The Heterogeneity of Catalyst Surfaces for Chemisorption HUGH S. TAYLOR Frick Chemical Laboralgl, Princeton University, Princeton, New Jersey
CONTENTS I. Introduction . . . . . . 11. Heterogeneity on Oxide Surfaces . 1. Zincoxide. . . . . . 2. Otheroxidea . . . . . 3. Interpretationof the Data . . 111. Heterogeneity on Metal Surfaces . 1. Copper . . . . . . 2. Iron . . . . . . . 3. Nickel. . . . . . . IV. Evidence from the Earlier Literature . 1. Nickel. . . . . . . 2. Iron Synthetic Ammonia Catalysts 3. Tungsten . . . . . . 4. ChromiumOxideGel . . . 5. Zinc-Chromium Oxide . . . 6. Zinc Oxide. . . . . . V. General Observations . . . . References . . . . . .
.
Page
. . . . . . . . . . . . . . . . . . . .
.
. . . . . .
.
. . .
.
. . . . . . . . ,
. .
. .
. . . . . . .
. . . . .
.
. . .
. . . .
.
. . . . . .
,
.
.
.
. .
. ..
. . . . . . .
. . .
.
. .
. .
. . . . . . . .
. . . . . . . . . . . . . . . .
.
. .
.
.
.
. .
.
.
.
.
. .
.
.
. .
. . .
. . . .
1 3 3 6
6 9 9 11 13 19 19 21 21 22
23 23 25 26
The technological development of catalytic processes was already well advanced at the conclusion of World War I but the basis for such development was largely empirical. No large body of scientific advance in the theory of catalysis had emerged in the interval between Faraday’s researches on the “power of metals and other solids to induce the combination of gaseous bodies,” in 1834, and the contributions of Langmuir (1)to the kinetics of surface action which originated in his studies of the “clean-up” of gases in a tungsten filament lamp.
I. INTRODUCTION Elsewhere (Taylor, 2) it was possible to outline the main features of the development of the basic scientific research that occurred between the two World Wars. These divided themselves naturally into four distinct periods of advance each of approximately 5 years duration. U p to 1925 research was concerned largely with the evaluation of the Langmuir concept of unimolecular adsorption of gases at the surfaces of catalysts and the consequent kinetic treatment of surface reactions. Between 1925 and 1930 1
2
HUGH 8. TAYLOR
attention w a devoted largely to the exploration of the heterogeneity of catalyst surfaces, to the importance of “active centers” on actual catalysts and to the consequences of such centem in terms of the features of poisons and promoter action already familiar in the industrial applications of catalysis. By 1930 it waa possible to differentiate between types of association between gas and catalyst Burface, to distinguish, in many cases, between nonspecific, physical or van der Waals adsorption of gases which has negligible significance in the phenomenon of catalysis and, on the other hand, a chemical association between gas and surface, chemisorption, involving chemical valency forces between surface atoms and adsorbed species and requiring for the formation of the adsorption complex an activation energy of adsorption. It was in this period, also, that new tools, notably the use of para-hydrogen and the deuterium isotope, were being increasingly employed to elucidate the nature of the interaction between gas and surface and between adsorbed reactant species on such surfaces. In the final years preceding World War 11,scientists in the field of eurface catalysis had succeeded (Emmett and Brunauer, 3) in obtaining a measurement of the accessible surface area of catalytic materials, reproducible in different laboratories and had essayed the task of exploring the specific activity of particular crystal faces in such catalytic materials aa nickel, iron, platinum, palladium and rhodium (Beeck, Smith, and Wheeler, 4) which could be produced in film forms with particular oriented faces of the crystal predominantly present in the catalyst preparation. During the war years there is definite evidence that fundamental research in the area of contact catalysis went into eclipse. It is not yet possible to dikern the direction in which new developments of the most significance will occur. We can, however, call attention to programs which tend to renew interest in the problem of heterogeneity of actual catalyst surfaces, which formed an important phase in the second five year period already mentioned and which was in sharp contrast to the attitudes prevailing immediately prior to the war. Then, there waa a definite tendency to minimize all emphasis on the imperfections of crystal structure in catalysts, the distortion of faces, abnormal dimensions in the crystal lattices, extralattice structures resembling more nearly the gaseous species of an element. Rather, the properties of a catalyst were being interpreted in terms of the normal faces of particular crystals and their statistical distribution in a particular catalyst preparation. The method of evaluating the total surface area of catalyst preparations that resulted from the researches of Brunauer, Emmett, and Teller (5) on the van der Waals adsorption of gases, such as nitrogen and butane, at temperatures near to their respective boiling points, provided a very powerful tool with which to re-examine the whole problem of “active centere” 01
CATALYST S m F A C E S FOR CHEWSORPTION
3
the heterogeneity of practical catalysts. In the period 1925-1930,when research into such matters was most intense, the scientist had no adequate measure of the extent of surface in the preparations which he studied. Intercomparison of different preparations was therefore severely limited and restricted. Nor could the extent of active surface, presumed from the studies to be present, be estimated at all in terms of total area accessible to gas. The development of the Brunauer-Emmett technique changed this condition materially for the better. It became possible for the scientist to define rigorously the extension of a surface which his particular method of preparation of a catalyst would produce and to state his conclusions concerning a particular measurement of an active area in terms of a total avail,able area. The techniques were at hand to define a percentage efficiency of a total specified surface. One other factor contributed to the desirability for the re-examination of the concept of active centers. The development of a distinction between van der Waals adsorption and chemisorption only occurred after the interest in heterogeneous surfaces had somewhat waned. A return to the study of heterogeneity of surfaces in the postwar years could be assisted by the sharper differentiation between the two types of adsorption that resulted from the development of the concept of and criteria for chemisorption. As an illustration of the possibilities inherent in such studies record will be made of a group of researches now under way in the Princeton laboratories in which both measurement of surface areas and differentiation between van der Waals and chemisorption are being utilized for a restudy of the variation in the character of different catalyst preparations. Work already accomplished has permitted an evaluation of a aeries of oxide catalyst preparations such as have found use in catalytic hydrogenations, notably zinc oxide, chromium oxide, zinc-chromium oxide mixtures, and manganese-chromium oxide preparations. The methods of study are presently being employed in corresponding researches on the heterogeneity of metal catalysts, notably copper, iron, and nickel. The results already accumulated suggest that interesting generalizations and significant applications will follow. 11. HETEROGENEITY ON OXIDESURFACEB 1 . Zinc Oxide General Considerations. It was decided to re-examine the adsorption of hydrogen at various oxide surfaces. Earlier work (Taylor and coworkers, 6), in the years 193G1938, had indicated certain advantages from such choice. In many such catalyst preparations the temperature regions in which van der Waals adsorption and chemisorption of hydrogen a.
4
HUGH 8. TAYLOR
occur are sharply separated one from the other due to a relatively considerable energy of activation of chemisorption on such oxide surfaces. Thus, the earlier work had revealed (Taylor and Sickman, 6b) that on zinc oxide surfaces the chemisorption of hydrogen set in measurably at temperatures above 0" C. whereas in this temperature range the van der Waals adsorption of hydrogen must be negligibly small. A result, obtained in that earlier work (Taylor and Strother, 6c) provided an additional incentive for a restudy of einc oxide preparations. Zinc oxide, prepared from the oxdate, had shown in the region of chemisorption two maxima of chemisorbed gas at temperatures in the neighborhood of 80 and 218" C. With the exception of two similar maxima observed by Emmett and Harkness (7) in the chemisorption of hydrogen on iron synthetic ammonia catalysts, and labelled by them type A and type B chemisorptions, these maxima in the zinc oxide adsorptions of hydrogen were without parallel. The interpretation of two types of chemisorption of hydrogen on a catalyst surface has not, thus far, been satisfactorily formulated. It was, therefore, decided to restudy the whole problem of hydrogen adsorption on zinc oxide, ex-oxalate, securing simultaneously the pertinent Brunauer-EmmettTeller (B.E.T.) data as to total accessible surface on the preparations ennmined. 21. Technique. The method of study chosen was to meaaure a t mccessively increesing temperatures from - 195' C.(liquid nitrogen temperature) to 302" C. the extent of adsorption of hydrogen, the working pressure being maintained continuously at about 1 atmosphere pressure (Taylor and Liang, 8). In the temperature range below 0' C. the measurements revealed a steadily decreasing adsorption with increasing temperature as would be anticipated for van der Waals adsorption. Above 0" C. a phenomenon of considerable importance became evident. In thie temperature range where van der Waals adsorption ie negligible it waa found that, in passing from one operating temperature to another, higher temperature, marked desorption of gas occurred, to be followed after a measurable interval of time by a readsorption of the gas. The desorption process was relatively rapid, the readsorption was a typicd rate process of chemisorption. An example of such an observation is plotted in Fig. 1. The diagram further indicates that after the rate of adsorption, subsequent to the change of temperature from 111 to 154°C.had become negligibly slow, a return to the lower temperature, 111" C., resulted in the rapid readsorption of gas in an amount corresponding closely to that which had been desorbed when the temperature was elevated through the given inte.rval. This observation suggests immediately that, during the tempers ture rise, gas has been desorbed from a given area of the surface on which it waa held (by chemisorption, it can be shown) at the lower temperature.
5
CATALYST SURFACES FOR CHEMISORPTION
The surface laid bare by the desorption apparently remains bare at the higher temperature, although adsorption proceeds at this higher temperature, presumably therefore on other areas of the total surface. When the
a: * 6 v)
V 0
0
w 5
m
a 0
v)
0
a 4 W
5
3 J
z 3
so
350
650
950
TIME IN M I N U T E S
1250
Fro. 1. Adsorption rate curve on raising and lowering temperature (hydrogen on ZnO I).
temperature is lowered, the bare surface immediately refills with chemisorbed gaa at a rapid rate in agreement with the observation that it was rapidly desorbed when the temperature was elevated. This phenomenon of rapid desorption on raising the temperature followed by a slow readsorption on other areas was consistently observed
ieo
300
400
600
600
TEMPERATURE,' K
FIG.2. Adsorption isobar (hydrogen on ZnO 111). Roman numeral indicates order of run. 0 : Raising temperature A: hwering temperature
6
HUGH 6. TAYLOR
in the case of zinc oxide chemisorption at temperatures between 56 and 218" C. This led to an examination of the behavior of the preparation as it was cooled in hydrogen at 1 atmosphere pressure from 302" C. to liquid nitrogen temperatures. An isobar for zinc oxide on raising and lowering the temperature through these ranges is shown in Fig. 2. The increments of adsorption on lowering the temperature through a given interval roughly paralleled the rapid desorptions that occurred as the temperature was raised through the same interval. That the adsorptions involved were actually due to chemisorption was indicated by measurements of nitrogen adsorption on the zinc oxide surfaces studied. These measurements showed that already a t 56' C. the van der Waals adsorption of nitrogen was too small to measure on zinc oxidc surfaces with surface areas by the BET method in theneighborhood of 5-25 sq. meters/g. The van der Waals adsorption of hydrogen would be still smaller and lower by one or more orders of magnitude than the quantities of gas involved in the desorption and readsorption processes.*
3. Other Oxides These observations, when extended to other oxides or mixtures of oxides, gave in part confirmatory data and, in one case, that of manganesechromium oxide, data which significantly differed from those obtained with zinc oxide. Thus, with chromium oxide gel with a surface area of 189.5 sq. meters/g. the quantities of gas involved in the desorption-readsorption phenomena between 56 and 302" C. amounted to 9 cc. or nearly 5% of the total surface. With a zinc chromium oxide of 21 sq. meterdg. the aren involved in desorption-readsorption phenomena was 13% of the total surface between 0 and 302" C. With manganese-chromium oxide, on the other hand, no readsorption of gas was observed when the sample was cooled in hydrogen from 218 to 0" C. 3. Interpretation o j the Data A schematic representation of these several conditions can be seen from Fig. 3. If the continuous lines A B and CD represent true equilibrium isobars for van der Wads and chemisorption respectively, it is obvious that, on raising and lowering the temperature in the CD region the same experimental values will be obtained. If the dotted line BC represents values obtained while the temperature is being raised it is obvious that these are below the true equilibrium values because the latter cannot be reached in practicable times of observation since the velocity is too low. If the
* Thls ariaes from the identity of the van der Waala forces between gee molecules and the van der Waals forces between gas molecules and EUl'faCCE resulting in stronger edsorptione of the more easily liquefiable gas.
CATALYST SURFACES FOR CHEMISORPTION
7
surface were so uniform or the hydrogen held so strongly that no desorption had occurred in the ascending portion BC the adsorption data which would be obtained on cooling below the point C would be hmkontal until the van der Waals adsorption set in again. This is indicated by the horizontal section CE and the ascending section EF. This is the case to which mane
TEMPERATUAE,'K
FIG.3. Schematic representation of chemisorption on lowering temperature with uniform and heterogeneoussurface.
ganous-chromium oxide conforms, for the probable reason that the chemisorption of hydrogen by this surface is strong with a heat of adsorption of ca. 20 kcal./mole and an activation energy of 10 kcal./mole (Taylor and Williamson, 6a). The rate of desorption with an activation energy of -30 kcal./mole would be small in the ascending section BC of the curve. When, however, desorption does occur in the ascending portion of the curve BC, there should, on cooling down from C,no longer be a horizontal section C E but continuously increasing values of the adsorption CG, measuring, in excess of the values over the horizontal section, the quantity of gas which was desorbed during the measurements made in passing upwards in temperature from B to C. It is this behavior which is shown by zinc oxide, chromium oxide gel, and zinc-chromium oxide. This aspect of heterogeneity of surface which the desorption-readsorption phenomena reveal is worthy of a more detailed analysis. Consider, for example, a hypothetical adsorbent, having two distinct adsorbent areaa A and B, characterized by two distinct activation energies EA and E B , and two characteristic heats of adsorption XA and XB. It is possible to conceive two temperatures TI and Tz such that, a t the lower temperature
8
HUGH 0. TAYLOR
TIonly the area A will be filled while I3 remains sensibly bare, whereas at the higher temperature T2the area A will be bare and area B w illbe sensibly covered. These two possibilities are illustrated in Fig. 4. The requisite
€ 1 ,
A i i
Eat
A c i
El,
i l
i
El.
kc;
Fxa. 4. Hypothetical adsorbent with two areas of different characterktics.
conditions are the following. If, at TI,the magnitude of the activation energy EA is sufficiently low and XA sufficiently high the area A will be progressively covered with time. The condition for B to remain sensibly bare is that E B be sufficiently large so that negligible chemisorption occurs due to the smallness of the Boltzmann factor e - E B ’ R T L in the equation for the velocity of chemisorption. For the reversal of this condition at temperature Ta the necessary conditions are, (1) that the magnitude of ( E A + X A ) , the energy of desorption from area A , shall be low enough to permit rapid evaporation from this area, and (2) that EB shall be of such a magnitude that chemisorption shall occur with measurable velocity, and ( E B + XB) be sufficiently large that the surface shall be progressively covered. Given such conditions it would obviously be possible to select a working pressure P such that at TI the fraction B of a given surface covered will approach 0 ~ = 1while 0,-0; at T2,OA-Oand 0, approaches unity. Furthermore, at such a pressure P , the composite adsorbent A.13 saturated on the area A at TIwould, on changing the temperature to T,,suffer desorption of gas from area A and adsorption eventually would reach saturation on B. If the desorption process were more rapid than the readsorption process the observer would record, on raising the temperature, first an evolution of gas and then a readsorption. It is this phenomenon which is characteristic of the result already outlined with oxides and with some metals as next to be recorded. It should be emphasized that with all actual adsorbents the conditions will never be as simple as those discussed in the hypothetical adsorbent. The more complex surfaces will show a more complex pattern of behavior, actually an integration of both desorptions and adsorptions, each with its own characteristic velocity on a particular surface.
Activdion Energy of Chemisorption on Heterogeneous Surfaces. A further consequence of a heterogeneity of surface such as is indicated in
CATALYST BURFACEB FOR CHEMISORPTION
9
Fig. 4 has also been pointed out by Taylor and Liang (8a). Measurement of the velocity of adsorption of a given gas on a clean surface AB at temperature TIwould record the amount of gas adsorbed with time on area A only. At temperature Ta the measurement would record the rate of adsorption on area B. The two velocities would actually have no relation to each other. They should not be used to determine activation energies by means of the equation
where ul and uz are the measured velocities at the two temperatures. Because the processes at each temperature occur on different areas they actually have no connection with each other. They cannot be used for determining the activation energy of a process because they are two distinctly different processes. Conversely, if from velocity data at three or more temperatures strongly varying values for the activation energy of adsorption of a given volume of gas are obtained, the data are direct evidence of the heterogeneity here under analysis. The early data of Taylor and Strother (6c) reveal this, while the later data of Liang (8b)are much more emphatic evidence of the same condition in the case of zinc oxide.
111. HETEROGENEITY ON METALSURFACES 1. Copper In Figs. 5 and 6 are reproduced data for copper catalysts which show desorption-readsorption phenomena and also the variation in a hydrogen adsorption isobar on raising and lowering the temperature. Catalyst B-I of Fig. 5 was a copper-magnesium oxide catalyst obtained by coprecipitation of the oxides and reduction of the copper to metal in hydrogen at 350" C. The adsorption of hydrogen waa qualitatively similar to that obtained by Lewis and Taylor (9). The data reveal the desorption-readsorption phenomena at 56 and 302". Catalyst B-4 was prepared from a nitric acid solution of 63 g. copper containing less than 0.0005% nickel but with the addition of C.01 mole thorium nitrate. The solution was evaporated to dryness and ignited to oxide, after which reduction was effected at 154" C. The thorium nitrate served to maintain some extension of the surface. A pure sample of copper nitrate ignited and reduced at 280" C. gave a surface area of 0.04 sq. meters/g. A pure sample reduced at 110" C. over a period of 10 days gave an aces of 0.74 sq. meters/g. With the addition of 0.01 mole thorium nitrate/mole of copper the surface, after reduction at 154" C., waa 2.86 sq. meters/g.
10
HUGH 8. TAYLOR 30
0'
2
8 25 d w m
a 0
v)
0
a w 20
I 3 J
0
>
I5
I
I
0
500
1000
I
I
I
1500
2000
2500
I
I
3000
3500
I
4060
TI ME, MINUTES
FIQ.6. Hydrogen adsorption rates on copper catalysts at 1 atmosphere.
70
I IS0
I
200
I
250
I
300
I
350
TEMPERATURE,.K
FIG.6. Hydrogen adsorption isobar on copper B-4 at 1 atmosphere. Vm-13.f3 cc. 0:Lowering temperature
A: Raising temperature
11
CATALYST SURFACES FOR CHEMISORPTION
The data illustrated in the two diagrams are evidence that the same phenomenon of desorption and readsorption is found with copper catalysts, either supported or essentially pure metal, as is found with the oxides. 2. Iron
With doubly-promoted iron synthetic ammonia catalysts the same type of observations are repeated. Data on desorption and readsorption are shown in Fig. 7, while in Fig. 8 are the values obtained on raising and lowering the temperature. It will be noted that the latter curves do not show the double maxima at -78 and 110" C. earlier found by Brunauer and Emmett (10) and examined by them in detail as types A and B adsorption. A T 0 53.3
75
CC.
AT 78.3.K
I
I
I
zoo
500
400
1 500
600
TEMPERATURE,'K
FIQ.7. Isobaric hydrogen adsorption on iron CF 413-9at 1 atmosphere. 0:Raising temperature
A: Lowering temperature
The present curve does not show the maximum found by the previous authors at -78°C. Private communication from Dr. Paul Emmett indicates that the possible explanation for this is that the reduction procedure here employed may have failed to reduce the iron in those areas where the low temperature chemisorption of hydrogen occurs. According to Emmett these areas are only uncovered by reduction at around 500" C. in a rapid stream of hydrogen freed from oxygen and well dried before passing into the reduction zone. The data in Fig. 7 do, however, shorn that from room temperature upwards there is alwa- s evidence of desorption on raising the temperature followed by readsovtion, and the
HUGH B. TAYLOR
12
data shown in Fig. 8 for lowering temperature serve to confirm these observations. The iron catalyst used was a technical synthetic ammonia catalyst containing only 1-2% of the potash-alumina promoter.
::y y i-1:1 . L9
c o
2000
N O
1000
IS00
2600
3100
3600
4100
4600
SBOO
6300
6800
7300
7800
m
a
v) 0
10
0
a
BOO0
T I M E I N MINUTES
Fro. 8. Rate of hydrogen adsorption on iron CF 413-9 (Vm-67 cc.) a t 1 atmosphere (18.4 cc. adsorbed a t 24" in 8 days before A). First reading at: A, 56"; D, 154";F, 218'. Change temperature at: B, 56 to 111"; C, 1 1 1 to 156'; E, 154 to 218".
ii
d W
(I-
m
a
0
v) Q 4-
a
W
f-1 0
> 75
I
1
200
300
I '
400
1
so0
i 600
TEMPERATURE..K
FIQ.9. Isobaric hydrogen adsorption on iron 931 a t 1 atmosphere. Vm-7.58 0 : Raising temperature A: h w e n n g temperature
CC.
CATALYST SURFACES FOR CHEMISORPTION
13
In Fig. 9 are shown comparable data with a small sample of the Fixed Nitrogen Laboratory Catalyst No. 931. Here there is some evidence of a maximum around -78" C. The curve for lowering temperature is evidence for the phenomenon of desorption-readsorption.
3. Nickel The most extensive experiments on the behavior of nickel catalysts have been carried out by Mr. H. Sadelc. The results will appear in his forthcoming dissertation and later in the scientific journals. The results may here be summarized. Since the researches of Benton and White (11) it has been well known that chemisorption occurs on technical nickel catalysts a t temperatures only slightly above liquid air temperatures. On nickel films of the type produced by Beeck (4a) it is well known that even at liquid air temperatures the hydrogen adsorption observed is chemisorption. For these reasons Sadek has made a particular study of the adsorption of hydrogen on a variety of nickel catalysts prepared by usual technical catalyst techniques and has examined them in particular detail in the temperature range from -195 (liquid nitrogen) to -78" C. (COa point). He has found that adsorption studies can be carried out at the melting points of methylcyclohexane and toluene, i.e., a t -126 and -95" C., where the rates of chemisorption are especially interesting with some of the nickel preparations studied. a. Nickel Kiesdguhr. In Fig. 10 are shown measurements made on a nickel-kieselguhr catalyst (15% Ni) from a large batch prepared for us many years ago by the E. I. du Pont de Nemours Company and which has been used in many earlier Princeton studies of the catalytic activity of nickel, notably by Morikawa (12) and more recently by Kemball (13). The sample employed in the present studies was reduced at temperatures up to 400" C. and examined in the temperature range -195 to 302" C. The sample used for the work shown in Fig. 10 had a surface area measured with nitrogen equal to a monolayer of 50 cc., i.e., 85.5 sq. meters/g. on the 2.55-g. sample. Figure 10 shows the isobars obtained starting with a clean surface a t the several temperatures of - 126, -95, and -78" C. and separate isobars representing the data obtained 10, 60, 120, and 250 minutes after hydrogen a t 1 atmosphere pressure was introduced to the evacuated sample. It will be noted from the 10-minute isobar that already at - 126" C. the van der Waals adsorption of hydrogen is low. Actually it must be less than the value after 10 minutes if any chemisorption has occurred. The van der Waals adsorption of hydrogen must be still lower than the - 126" C. value a t the higher temperatures of -95 and -78" C. Nevertheless the 10-
14
HUGH 8. TAYLOR
minute readings at these temperatures are much higher than at - 126" C., which indicates that at these temperatures chemisorption haa occurred more rapidly than at - 1 2 6 O C. Chemisorption is occurring in these temperature intervals at measurable rates and this is illustrated by the rate
TEMPERATURE,. K
FIG.10. H2 Adsorption isobnr of nickel kicselguhr (DuPont). Vm=50 cc.
data in Fig, 11. These rate data are in every way similar to those obtained with zinc oxide and other oxides in the temperature regions of 56 to 302" C. They are measurable at - 126 to -78" C. because the activation energy is lower; but, nevertheless, an activation energy is required and determines the rate process. Especial attention should be directed to the lower of the two isobars represented by the full lines drawn through the experimental points from - 195 to 302' C. This was obtained by making first a measurement at liquid nitrogen (- 195" C.) and, when saturation was apparently reached, changing the temperature successively to -126, -95, -78, 0, and so on until the temperature of 302' C. was reached. "his curve is analogous to those in the preceding cases of the oxides and metals designated as points on "raising" the temperature. The upper curve represents the data obtained on lowering the temperature. It will readily be seen that between 0 and 302" C. the true equilibrium isobar is approximately reached. Below 0" C., howel-er, the equilibrium isobar is not attained by the technique of
15
CATALYST SURFACES FOR CHEMXSORPTION
"raising" the temperature while maintaining constant atmospheric pressure. It is interesting to observe parenthetically that, for example, at -126" C. the volume of hydrogen adsorbed on a clean surface even after
$12
t
v)
Y
10
n W
0 v)
0 - DRY ICE
2 6 W
5
-I
0 - MELTING TOLUENE 4
0 - MELTING METHYLCYCLOHEXANE
0
>
2
I 20
100
I
I
I
200
300
400
TIME IN MINUTES
FIG.11. Rate o f adsorption of Hz on clean surface of n i c k e l k i e s e l g u h r (Du Pont). V, =50 cc.
250 minutes (and even after 400 minutes as the rate curve of Fig. 11 shows) is markedly lower than the value obtained after only 150 minutes when the system is brought up to - 126"C. from liquid nitrogen temperatures. This is not an isolated example. There are numerous similar data in other surfaces studied. The nickel-kieselguhr surface is covered with hydrogen to an extent equalling nearly one-half a monolayer of nitrogen at liquid nitrogen temperatures. The high value recorded at -126" C. on raising the temperature compared with that obtained after 250 minutes on a clean surface at -126" C. suggests that, during the interval of time as the temperature is rising from - 195 to - 126' C., much of the van der Waals adsorbed hydrogen is converted to chemisorbed gas and held by the surface in this manner. As already observed, the extent of van der Waals adsorbed hydrogen at - 126" C. cannot exceed that recorded after 10 minutes on a clean surface. These observations are indicative of the truth of an assumption frequently made in the Princeton work that the process of chemisorption involves an activation energy effect on a surface-van der Waals adsorption complex. Actually, in 1932 it was pointed out (Taylor, 14) that on any
16
HUGH 6. TAYLOR
reasonable assumption concerning the extent of zinc oxide surfaces, with activation energies of the magnitudes observed (10-15 kcal.) the velocities of chemisorption were entircly too low to be explained on the assumption that the rate of adsorption was determined by the number of gas molecules striking the surface per unit of time and having the required activation energy in the form of kinetic energy. The present data support an assumption of an activation of a surface-van der Waals adsorption complex. b. Sintered Nickel-Kieselguhr. Returning to the problem of heterogeneity in such nickel surfaces it is calculable from the rate data in Fig. 11 as well as from a comparison of the curves obtained by raising and lowering the temperature in Fig. 10 that, within the temperature range -126 to 0’ C., the principal heterogeneity of the nickel surface is observed. This is confirmed by the measurements recorded in Fig. 12 for the same nickel-
TEMPERATURE? U
FIG.12. Adsorption isotmr of TI, on “sintered” nickel kieselguhr (Du Pont).
vm=33.77 cc.
kieselguhr catalyst after it had been “sintered” by heat treatment for 24 hours at 500” C. This sintering resulted in a loss of surface areaindicated by the new nitrogen monolayer value of 33.77 cc., a decrease of 33% in the available surface. In Fig. 12, the van der Waals adsorption of hydrogen has decreased below that in Fig. 10 by a similar amount. At -126” C.,
CATALYST SURFACES FOR CHEMISORPTION
17
on the "rising temperature curve," the minimum of adsorption observed a t - 126" C. amounts t o some 20% of that obtained a t the same temperature on the unsintered catalyst by the rising temperature technique. This means that the chemisorption requiring only low activation energies has been suppressed by the sintering operation. There is a notable decrease in the low activation energy areas. The divergence between the "rising temperature curve" and the lowering temperature curve sets in, not a t about 0" C. as with the unsintered sample but somewhere between 0 and 110°C. Figs. 10 and 12 effectively illustrate a redistribution of the heterogeneity of the surface brought about by the process of sintering by heat treatment.
_-----
r
0-ICE
)
0 - MELTINQ TOLUENE
1
0 - DRY ICE
I
I
I
I
20
60
100
I50
TIME IN MINUTES
Fro. 13. Rate of adsorption of H2 on clean surface of (Ni -Tho,2%). 8, =29.82 cc.
Nickel-?% Thoria. Changes in the distribution of heterogeneity on nickel surfaces can also be secured by differing techniques of preparation. Two examples in illustration of this may be cited from among a number of such cases which wilI be detailed in the chemical journals. A preparation of nickel containing 2% thoria ae promoter, obtained by evaporation of the nitrate solution calcination to oxide and reduction in hydrogen, was more active for chemisorption of hydrogen in the low temperature range than was the nickel-kieselguhr unsintered preparation already discussed. The rate data shown in Fig. 13 show a more rapid rate of chemisorption of hydrogen a t - 126" C. than in the case of nickel-kieselguhr. At -95" C., the rate of c.
HUGH 0. TAYLOR
18
chemisorp,tion is so rapid that saturation is reached within 40 minutes of the start of the determinations, in contrast to the data at this temperature shown in Figs. 10 and 11. The contrast in disposition of the rate curves at the temperatures -126, -95, and -78" C. in Figs. 11 and 13 is definite evidence of the different distributions of heterogeneity in the two nickel surfaces. With the nickel-thoria catalyst there is definite evidence also that at 0" C. new areas are opened up to chemisorption of hydrogen not available at the lower temperatures. d . NickeUlhromia. In marked contrast to the three preceding examples of nickel catalyst is the example of a nickel-chromia catalyst containing 80% nickel and 20y0 chromium oxide. This catalyst, which has been found to have desirable hydrogenation characteristics and some measure of technical application, is entirely different in adsorption characteristics from the preceding examples. The curves in Fig. 14 suggest at
12
I 300
400
500
TEMPERATURE ,*K
FIG.14. €I, Adsorption isobar on Ni-CrzO,.
Vm-3G.38 cc.
first that this catalyst exhibits chemisorption of hydrogen only in the temperature range above 0" C. It is more like a chromium oxide gel catalyst than the nickel catalysts already considered. It reveals, in the form of the curve obtained on lowering the temperature from 302 t o 0" C., that extensive desorption-readsorption occurs in this temperature range when the system is subjected to the technique of raising the temperature. The irregular form of the rising-temperature curve is further evidence of this. The large quantities of gas adsorbed a t liquid nitrogen and dry ice
CATALYST SURFACES FOR CHEMISORPTION
19
temperatures in terms of the nitrogen monolayer suggest, however, that some of this adsorbed gas is chemisorbed and at a rapid rate. There appear to be in this catalyst at Ieast two main temperature ranges where hydrogen is chemisorbed, below -78" C. and above 0" C., a definite index of wide variations in activity of the catalyst for chemisorption.
IV. EVIDENCE FROM
THE
EARLIER LITERATURE
1. Nickel
Two sharply differentiated types of active center for chemisorption of hydrogen on nickel were revealed by calorimetric and desorption studies conducted by Eucken and Hunsmann (15). The differentiation between van der Waals and chemisorption by means of calorimetric data was made convincing by calorimetric measurement of the heat of adsorption a t temperatures from 20 to 195" K. At 20" K. physical adsorption only was observed with a mean heat of adsorption of 800 calories/mole. At 55" K. the adsorption was, in part, chemisorption since the observed heat of adsorption was somewhat less than 5 kcd./mole. At the three temperatures 90, 195 and 273" K. where the first quantities of gas were adsorbed with approximately the same heats of adsorption, in the neighborhood of 20 kcal./mole the gas appears to be completely chemisorbed. The catalyst wm prepared by ignition of the nitrate to oxide and was carefully reduced in hydrogen at 300" C. for 4 days. It would appear to be more active for chemisorption than the preparations discussed in the preceding sections judged by the low temperature, 55"K., at which chemisorption was calorimetrically evident. On the other hand, the authors' statement that a 50-g. ming of nickel in the calorimeter absorbed only 40 cc. hydrogen at 0" C. would be indicative of a low adsorption capacity in terms of the preceding data where adsorption at 0' C. and 1 atmosphere pressure attained, in the case of nickel-kieselguhr, 19.6 cc./g., and with the n i~k el-2 7thoria ~ reached 1.4 cc./g. The desorption technique devised by Eucken and Hunsmann consisted in saturating the nickel preparation with a quantity of hydrogen, rapidly pumping off the dead space and easily desorbable gas at temperature, and then heating the preparation for 1-5 hours up to 600" K. with continuous evacuation, At regular intervals the temperature and hydrogen pressure over the nickel were observed and recorded. The data show that with adsorption a t 195 and 273°K. there are two stages of desorption, one practically complete at 273" K., the other first measurable at 473" K. The first stage involved 14% of the gas, the remainder being desorbed at the higher temperature. The gas adsorbed at 20" K. could be desorbed completely at slightly higher temperatures than 20" K. Also with adsorption
20
HUQH D. TAYLOR
at 83"I<. the adsorption was practically completely reversible at temperature 80 that the authors decided that it was not conclusive that there waa a conversion a t this temperature from physical to chemisorption. Nevertheless, the calorimetric data are decisive that at 90" K. the gas is chemisorbed. The conclusion seems compelling that the energy of desorption of hydrogen chemisorbed at 83°K. is markedly less than the energy of desorption of hydrogen chemisorbed a t 195 and 273" K., again evidence for the heterogeneity of catalyst surfaces for chemisorption. Eucken and IIunsmann, from their calculated energies of desorption and their measurements of heats of adsorption, construct:d a chart for the distribution of active centers over their nickel catalyst surface. Their graph is reproduced in Fig. 15. The horizontal Iines indicate the areas which are covered by adsorption a t the temperatures 90, 195, and 273" K. These are determined by the activation energies of adsorption, measured by Ed- AHa, where E d is the activation energy of desorption and AHa is the heat of adsorption.
1
25
GI..
.
attar bnq contact
0
o
I
I
I
I
I
I
I
10
20
30
40
50
60
70
I
I
eo 90
NUMBER OF ACTIVE CENTERS
loo
FIG.15. Distribution of activation energy of desorption ( E d ) and heat of adsorption AHa, of the cLct.ivespots on nickcl surfacc. (The horizontal lines give the regions which become covered at the indicated temperatures.)
Special at.tention should be directed to the horizontal lines for 273" K. indicating immediate coverage arid coverage after long contact. These are different. This arises firstly from the magnitude of the activation energies,
CATALYBT SURFACES FOR CHEMIBORPTION
21
the areas of lowest activation energy being first covered. On standing, however, gas molecules are desorbed and readsorbed by unoccupied centers of higher heats of adsorption. It is in this fashion that Eucken and Hunsmann reach an explanation for an observed heat evolution in the calorimetric measurements after the adsorption of gas is complete. “The heat must in large part be due to a slow process which accompanies the adsorption and is not demonstrable manometrically.” This is undoubtedly the process of desorption and readsorption occurring a t a single temperature which is so very pronounced in the Princeton t e c h i q u e of raising the temperature at constant pressure. It could only arise with a surface having heterogeneous character.* 2. Iron Synthetic Ammonia Catalysts Brunauer and Emmett (10) in their careful studies of iron catalysts had already observed (see especially Tables IV, JJ and XI11 of the paper cited) the phenomenon of increasing adsorption with decreasing temperature that we have indicated as occurring in our studies, as recorded in Fig. 8, on other preparations of synthetic ammonia catalysts.
3. Tungsten I n the studies of hydrogen and nitrogen adsorption on tungsten powders Frankenburger and his associates (16) established the fact that the adsorption of nitrogen on tungsten increased with increasing temperature, that the establishment of equilibrium was a slow process and that, on cooling the system, the amounts taken up on increasing the temperature are not given off again. In a footnote (see reference 4 of the paper cited), in the article by Messner and Frankenburger (16a) there is recorded a n observation essentially the same as that investigated in detail by Taylor and Liang. With a tungsten powder which had come to pressure constancy with nitrogen at a given temperature it was found that “after rapid temperature rise there is first a rapid increase in pressure, which is followed by a considerably slower decrease in pressure; the latter finally considerably outweighs the primary pressure increase.” The authors follow this with the
* Throughout this paper the stress has been laid on the heterogeneity of the catalyst surfaces under study. Some workers in the field would assign the heterogeneity to a variation in the geometrical arrangement of the adsorbed molecules on the surface. From this point of view i t might be of greater generality to speak of the heterogeneity of the surface-adsorption complexes. We believe, however, that new chemisorption data on a variety of surfaces, notably the measurements of Frankenburg ( J .A m . Chem. SOC., 66,1827,1838 (1944))of hydrogen on tungsten surfaces in which there is a linear relation between log 8 and log p , 6 being the fraction of the surface covered, cannot he explained by interactions between adsorbed molecules. They demand a heterogeneity of adsorption sites (SIX, for example, Halsey and Taylor, J . Chem. Phys., 16,624 (1947)).
22
HUGH 8. TAYLOR
remark: “according to our view these two pressure effects correspond to a desorption and subsequent chemical binding.” The data do not permit an estimate as to what extent the desorption occurring was of van der Waals’ or ehemisorbed nitrogen. If the latter, the evidence is direct that the tungsten surface is heterogeneous for nitrogen adsorption.
4. Chromium Oxide Gel In a study of the effect of gas pressure on the chemisorption of hydrogen on chromium oxide gel, Taylor and Burwell (6f) found it necessary to resort t o an arbitrary process of subtraction from experimentally observed amounts of adsorbed gas in order to display u relatively uniform area of gel for hydrogen chemisorption. Within close limits, a t temperatures of 457 and 491°K. and a t pressures of 1, 0.5 and 0.25 atmospheres, the energy of activation for the adsorption proccss was nearly independent of the amounts of adsorption ((t35 cc.) and the activation energy was about 21 kcal. In the tcmperaturc range 383 t o 457” K. no such uniformity was observed. For example, if the amounts adsorbed with time a t 405 and 427” K. be used to calculate activation energies the values received increase from 0 to 18.5 kcal. as the amount of grts adsorbed on the surface increases from 1.6 to 8.5 cc. Similar results can be drawn from the earlier measurements of Kohlschutter (17). As Burwell and Taylor observed, “the data as a whole appear to reveal (a) at higher temperatures, an activated adsorption with an activation energy and a pressure effect nearly independent of the amount of hydrogen adsorbed; and (b) a t somewhat lower temperatures, an activation energy and a pressure effect varying markedly with the amount adsorbed. “That any one definite process could be so changed in nature by a rather small change in temperature would be surprising. We will attempt to show that the deviation is but an apparent one, resulting not from a change in the nature of the higher temperature activated adsorption but from the increasing importance a t lower temperatures of a process nearly negligible at higher temperatures. “It will be assumed that some process causes the observed adsorptions throughout the greater part of the time-adsorption curve to exceed the adsorptions properly attributable to an adsorption with an activation energy of the order of 21 kcal. by the following amounts a t the different temperatures and pressuresJJ (see Table I). They go on to suggest that “the amounts of gas which we have been led to subtract from the observed data in the lower temperature range, 383.5-457” K., represent activated adsorption occurring with lower activation energies than that characterizing substantially all the adsorption above 457” K. Since the amounts subtracted (see Table I) diminish with increas-
23
CATALYST SURFACES FOR CHEMISORPTION
TABLE I Arbitrary Deductions From Observed Adsorptiona Temp..
O
383.5 405 42 7 457
K.
1 Atm.
Volumes, cc. 0.5 atm.
3.15 2.5 1.65 1.15
...
...
1.6 1.15 0.65
1.05 0.75 0.45
0.25 Atm.
ing temperature and decreasing pressure, and since only those sections of the curves after the eight, forty, and one hundred and fifty-minute abscissas at 427, 405 and 383.5’ K., respectively, are utilizable, we conclude that the subtractions approximate to the equilibrium values of the adsorptions with lower activation energies at the pressures and temperatures indicated.” The authors calculated that the heat of adsorption involved was of the order of 10 kcal./mole in contrast with a heat for the higher temperature adsorption of 27 kcal./mole. The desorption-readsorption observations of Taylor and Liang are confirmatory of this early interpretation of heterogeneity in chromium oxide gel surfaces. Beebe and Dowden (18) measured the heat of adsorption of hydrogen on chromium oxide gel at liquid air temperatures. They found a heat of adsorption of 5 kcal./mole which is larger than that to be expected for van der W d s adsorption. This gas, however, with the low energy of activation allowable a t liquid air temperature cannot be held to the surface in the temperature range 457 to 491°K. where chemisorption with an activation energy of 21.7 kcal. “on an area relat,ively uniform and large in extent” is occurring. 5. Zinc-Chromium Oxide As early as 1930 Garner and Kingman (19) reported that hydrogen adsorbed a t room temperatures and low pressures on ZnO-CrzOs catalysts waB given off a t 100-120° but after 20 to 30 minutes was readsorbed. At the time of these experiments it was not possible to state whether thegas adsorbed at room temperatures was van der Waals or chemisorbed gas. The later work by Liang definitely indicates that chemisorbedgas undergoes desorption and rcadsorption on such catalysts in the temperature region in question.
6. Zinc Oxide The heterogeneity of zinc oxide catalysts as revealed in the recent work of Taylor and Liang serves to confirm the conclusions of Taylor and
24
HUQH 8. TAYLOR
Smith (6g) reached by study of the velocity of interaction of Hzand D2 a t temperatures between 143 and 491" K. In that research it was shown t hut the catalyst operated in that temperature range with activation energies which increased from 0.6 kcal. in the temperature range 143-178" I<. through a region 195-373" K. with an activation energy of 7 kcal. to a high temperature region with an activation energy of 12 kcal. I n this latter there was a temperature range in which the apparent activation energy was zero. This latter abnormality was interpreted on the basis of a decrease in active surface covered with increase of temperature. A further observation of Taylor and Smith is pertinent. The velocities of the exchange reaction were occurring from 50 to 800 times faster than the net rate of adsorption of hydrogen a t atmospheric pressure, as measured by Strother on material of similar mode of preparation. The argument developed by Taylor and Smith was based on the postulate that the slow step in the exchange reaction was a desorption of chemisorbed hydrogen deuteride from the zinc oxide surface. Since this was occurring many times faster than the observed adsorption of hydrogen at a given temperature and pressure it would follow that the desorption process was occurring on an area of the surface where the activation energy of desorption was less than the activation energy of the adsorption process occurring under the specifred conditions of temperature and pressure. Two different areas of the surface were therefore involved, at a given temperature, in the exchange reaction and the adsorption process. The postulate of Taylor and Smith was challenged by others (20, 21) who were engaged on similar exchange studies on metal surfaces. These authors postulated a mechanism involving interaction between an impinging molecule and a chemisorbed atom attached to the surface. With this alternative postulate the argument of Taylor and Smith in favor of heterogeneity of surface could thus be discounted. The data of Taylor arid Liang which reveal the heterogeneity of such zinc oxide surfaces by experimental observations entirely free from any postulate with respect to reaction mechanism tend to strengthen the conclusions derived from the earlier experiments on exchange reaction. The great advantage of the study of exchange reactions of isotopic molecules on catalysts is that only one molecular specips is involved both as reactants and products. One is freed from the restrictions imposed with two reactants where the displacement of one reactant b y another or by a reaction product must steadily be taken into consideration. The catalysts which reveal heterogeneity by desorption-readsorption studies should show the same variation in activation energy of reaction with temperature that Taylor and Smith found with zinc oxide in the hydrogendeuterium reaetion. Research in this direction is under way.
CATALYST SURFACEB FOR CHEYISORPTION
25
V. GENERAL0 BSERVATIONS The heterogeneity of catalyst surfaces displayed in the desorptionreadsorption phenomena already detailed and in the reaction of isotopic molecules must be equally a factor in all catalytic reactions to a greater or less degree, dependent on the type of reaction and aIso the extent to which one or another reactant or product determines the reaction mechanism. In analyzing the process in any particular case it will be necessary to determine to what extent the total accessible surface, determined for example by the B.E.T. technique, is involved in the rate-determining slow process of the overall sequence. Thus, in the case of ammonia synthesis, where the activation of nitrogen is clearly the rate-controlling factor, it has further to be determined what fraction of the total accessible surface can play u role in such activation under a prescribed set of conditions. If similar conditions prevailed with nitrogen on synthetic ammonia catalysts as obtained with hydrogen on several of the catalyst surfaces here discussed then it mould follow that only certain fractions of the total accessible surface could be effective in a given temperature-pressure operation, and these could well be different in other reaction conditions. A further observation is pertinent. The present analysis forces the conclusion that, a t a prescribed condition of temperature and pressure, the slow processes of chemisorption or interaction with the surface by a reactant that can be demonstrated and measured at such a condition can have little significance in a rapid catalytic reaction that may occur under the condition specified. It is the processes of rapidly occurring desorption which are significant. To this extent the measurements of slow processes of chemisorption are not significant for catalytic reactions occurring rapidly a t the given temperature and pressure. They may well be significant for catalytic actions a t higher temperatures. For example, the slow process of interaction of carbon monoxide with Fischer-Tropsch catalysts to yield carbided catalysts, and measured in their progress over hours of observation, can have little significance in the catalytic synthesis of hydrocarbons which occurs at the same temperatures in time intervals of the order of seconds. It may well be, however, that the dcrnonstrated rapid interactions which occur in the first moments of interaction of carbon monoxide with the catalyst do possess significance. The nature of the interaction cannot readily be studied a t such a temperature owing to the rapidity of the interaction. Its occurrence in measurable time intervals a t a lower temperature may well have significance for the catalytic reaction at the higher temperature. Finally, it should be obvious that the variations in quality of surface that these studies of a particular group of catalysts have revealed go far to elucidating the variations in catalytic behavior of a catalyst that are com-
26
HUGH 8. TAYLOR
mon on examination, by different workers, of a particular mode of preparation of the catalyst. It is apparent that small variations in operational techniques can produce relatively large variations in the distribution of centers of activity. Indeed, variability in performance of catalysts produced from a given set of operations may well be the index of a pronounced heterogeneity of surface. REFERENCES 1. Langmuir, I., J . Am. Chem. SOC.38, 2221 (1916); ibid. 40, 1361 (1918); Trans. Faruday Soc. 17, 641 (1922). 2. Taylor, H. S., Am. Scientist 34, 553 (1946). 3. Emmett, P. H., and Brunauer, S., J . Am. Chem. SOC.69, 1553, 2682 (1937). London A177,62 (1940). 4. a. Beeck, O., Smith, A. E., and Wheeler, A., Proc. Roy. SOC. b. Beeck, O., Rev. Modem Phys. 17,61 (1945). 6. Brunauer, S., Emmett, P. H., and Teller, E., J . Am. Chem. Soc. 60,309 (1938). 0. a. Taylor, H. S., and Williamson, A. T., J . Am. Chem. SOC.63,2168 (1931). b. Taylor, H. S., and Sickman, D. V., ibid. 64, 602 (1932). c. Taylor, H. S., and Strother, C. O., ibid. 66,586 (1934). d. Taylor, H. S., and Howard, J., ibid. 116,2259 (1934). e. Taylor, H. S., and Ogden, G., Tram. Faraday SOC.80,1178 (1934). f. Taylor, H. S., and Burwell, R. L., J . Am. Chem. SOC.68,697 (1936). g. Taylor, H. S., and Smith, E. A., ibid. 60,362 (1938). 7. Emmett, P. H., and Harkness, R. W., J . Am. Chem. SOC.67, 1631 (1935). 8. a. Taylor, H. S., and Liang, S. C., J . Am. Chem. SOC.69,1306 (1947). b. Liang, 8.C., Ph.D. Thesis, Princeton Univcrsity, 1947. 9. Taylor, H. S., and Lewis,J. R., J . Am. Chem. Soc. 60,877 (1938). 10. Brunauer, S., and Emmett, P. H., J . Am. Chem. Soc. 02,1732 (1940). 11. Benton, A. F., and White,T. A., J . Am. Chem. Soc. 62,2325 (1930). 12. a. Morikawa, K., Benedict, W. S., and Taylor, H. S.,J . Am. Chem. Soc. 68, 1445, 1795 (1936). b. Morikawa, K., Trenner, N. R., and Taylor, H. S.,ibid., 69, 1103 (1937). 13. Kernball, C., and Taylor, H. S., J . Am. Chem. Soc., 70, 345 (1048). 14. Taylor, H. S., Tram. Faruday Soc. 26, 137 (1932). 15. Eucken, A., and Hunsmann, W., 2.physik. Chem. 44B,163 (1939). 16. a. hlessncr, G.,and Frankenburger, W., 2. physik. Chem. Bodenstein Festband, 605 (1931). b. Frankcnburger, W., and Hodler, A., Trans. Fardday SOC.28,229 (1932). 17. Kohbchutter, 11. W., Z. physik. Chem. 170A, 300 (1934). 18. Beebe, R. A., and Dowden, D. A., J . Am. Chem. SOC.60,2912 (1938). 19. Garner, W. E., and Kingman, F. E. T., Nature 126,352 (1930). 20. a. Eley, D. D., and Rideal, E. K., Nutwe 146,401 (1940). b. Eley, D. D., Proc. Roy. SOC.LondonA178,452 (1941). 21. Fmkas, A., and Farkas, L., J . Am. Chem. SOC.64,1594 (1942).
Alkylation of Isoparaffins V. N. IPATIEFF AND LOUIS SCHMERLTNG Universal Oil Produck Company, Rii!erside, Illinois
CONTENTS I. Introduction .
.
.
.
.
.
.
.
.
.
1. Historical . . . . . , . . . . 2. Scope . . . . . . . . , . . 3. Preliminary Survey . . . . . . . . 11. The Mechanism of the Catalytic Alkylation of Paraffins 111. Side Reactions . . . . . . . . . 1. Destructive .4lkylat.ion . . . . . . . . . . , . , , 2. Hydrogen Transfer . . . . . . . , , 3. Polymerization . . . . . 4. Formation of Catnlyst Complex . . . . . . . . 5. Formation of Esters IV. Catalysts . . . . . . . . . . . 1. Aluminum Chloride . . . . . . . . 2. Aluminum Chloride Sludges . . . , . . 3. Aluminum Chloride Monomethanolate . . , 4. Aluminum Chloride-Nitroalkane Solutions . , . 5. Aluminum Chloride-Alkali Metal Chloride Double Salts 6. Aluminum Bromide . . . . . . . . 7. Boron Fluoride . . . . . , . , . 8. Zirconium Chloride . . . . . . . . 9. Sulfuric Acid . . . . . . . . . 10. Hydrogen Fluoride . . . . . . . . References . . . . . . . . . . .
. .
. . .
.
.
.
,
,
,
.
. .
.
. . . .
.
.
. . .
.
. .
. .
,
.
. . . .
.
.
. . .
. .
.
. . .
.
. . . .
. .
. . .
. . . . . ,
Page . 27 . 27
. . . .
. . .
. .
. .
. ,
.
28
28 32 39 39 41 42 43 43 43 43 46 50 51
.
.
53
. . . . . .
. . . . .
53 54 54 55 59 63
.
I. INTRODUCTION 1. Historical The catalytic alkylation of saturated hydrocarbons with olefins was discovered and developed by Ipatieff and his co-workers in the laboratories of the Universal Oil Products Company (Ipatieff, 1). Experiments were carried out in June, 1932, by Ipatieff and Pines, using aluminum chloride as the catalyst, hydrogen chloride as a promoter, and hexane and ethylene as the reactants. These experiments having given positive results, they were repeated by Komarewsky, who then also investigated the alkylation of naphthenes. The alkylation of hexane was studied quantitatively by Grosse, who extended the reaction to other paraffins and catalysts, particularly boron fluoride. 27
28
V. N. IPATIEFF AND LOUIS SCHMERLING
Intensive research on catalytic alkylation of paraffins resulted in a number of commercial processes for the production of high octane number motor fuel from refinery gases. The reaction played a major role in World War 11, for it provided the means of producing “alkylate,” an important component of 100-octane aviation gasoline. For this purpose it had advantages over the alternative process involving olefin polymerization followed by hydrogenation because it yielded an isoparaffinic gasoline directly and because it permitted the utilization of butanes as well us of olefins, thus yielding approximately twice as much g a d i n e from a given amount of olefin. Alkylate was produced at thc rate of over five million gallons a day during the war. 2. Scope Only the results presented in papers appearing in the scientific chemical literature will be discussed. Papers dealing with the technical modifications and engineering fcatures necessary for the successful industrial application of the alkylation reaction to the production of gasoline are outside the scope of this article. Also, none of thc morc than seven hundred and fifty United States patents which relate to alkylation processes or to catalysts for them is to be considered here. It is the object of this review to present a critical discussion of the catalytic alkylation of isoparaffins and of the reaction mechanism. A comprehensive and detailed summary of the papers which had been published on this subject up to the middle of 1945, appeared in Chemical Reviews (Egloff and Hulls, 2).
3. Preliminary Survey The alkylation of paraffins with oldins to yield higher molecular weight branched-chain paraffins may be carried out thermally or catalytically. The catalysts for the reaction fall into two principal classes, both of which may be referred t o as acid-acting catalysts: (1) anhydrous halides of the Friedel-Crafts type and (2) acids. Representatives of the first type are aluminum chloride, aluminum bromide, zirconium chloride, and boron fluoride; gaseous hydrogen halides serve as promoters for these catalysts. The chief acid catalysts are concentrated sulfuric acid and liquid hydrogen fluoride. Catalytic alkylations are carried out under sufficient pressure to keep at least part of the reactants in the liquid phase. Propane and higher paraffins may be alkylated thermally. On the other hand, only paraffins which contain tertiary carbon atoms readily undergo catalytic alkylation. No method for the alkylation of methane and ethane in practical yield has been reported. The alkylation of isobutane has been most extensively investigated because the reaction converts this gaseous isoparaffin to liquid gasoline of high antiknock value.
29
ALKYLATION OF ISOPARAFFINS
Alkylation of isopentane, on the other hand, has been chiefly of academic interest since this isoallcane is already a liquid with a relatively high octane number. The gaseous olefins (ethylene, propylene, and the butylenes) are the preferred sllcylating agents. Low ratios of olefin to isoparaffin are used in order to minimize or prevent polymerization of the olefin. Mole ratios of 1 to 3 or 4 are usually employed in the laboratory; ratios as low m 1 to 8 or even lower are not uncommon in commercial alkylations. a. Thermal Alkylalion. The noncatalytic alkylation of paraffins will be given brief mention here primarily for purposes of comparison. Such alkylation requires high temperatures (about 500') and pressures (about 2500 t o 4500 p.s.i.). The main products of the reaction of ethylene with propane and imbutane, respectively, are pentanes and hexanes (Frey and Hepp, 3; Oberfell and Frey, 4). The reaction takes place by the addition of an alkyl group and a hydrogen atom derived from the alkane to the double bond of the olefin:
CHI-CHi-CHs
f CHT-CH~ +CHs-CHi-CHi-CHz-CHa
.1
CHs - CH - CHz - CHS
I
CH, CHa
I 1
CH,-CH+CHFCHZ+CH~ CHI
i
-CH -CHz -CHz-CIIs
I
CHI
CHs
I I
CHa - C -CHz -CHs CHo
A free radical chain mechanism is probably involved. example, CHI-CH - +CHs=CHi-+CHa - CH -CH*-CHz
AH1 CHI-CH -CHz-CHz-
I
CHa
I
Thus, for
-
CHIS +CHs-CH1 -CHs+CHs-CH
I
-CHz-CHs+CHa-CH
CHa
-
I
CH a
Since tertiary alkyl radicals are formed more readily than primary, it is to be expected that more isopentane than n-pentane will be obtained by the alkylation of propane, and more neohexane than isohexane will be produced from isobutane. The experimental results show this to be the case.
30
V. N. IPATIEFF AND LOUIS SCHMERLING
For example, the reaction of propane with ethylene at 510' and 4500 p.s.i. pressure resulted in a liquid product containing 55.5% by weight isopentane and 16.4% n-pentane, 7.3y0 hexanes and 10.1% heptanes; 7.4% alkenes were also present. The formation of alkenes and other by-products indicates that thermal cracking occurred. The alkylation of isobutane with ethylene under the same conditions yielded a liquid product containing 44.3% by weight neohexane, 11.6% isohexane, 1.1% n-hexane, 4,5'% heptanes, 9.6% octanes as well as minor amounts of other alkanes and alkenes. This reaction has served as a means for producing neohexane commercially. Thermal alkylation occurs readily with ethylene, less readily with propene and n-butylenes, and with difficulty with isobutylene. (As will be shown, the reverse is true in catalytic alkylation.) The reaction of propane with propene a t 505" and 6300 p.s.i. yielded a liquid product containing 18.070 2,3-dimethylbutane, 17.7% 2-methylpentane1 and 5.4y0 n-hexane. An even higher pressure, 8000 p.s.i. was required for the reaction of isobutylene with isobutane at 486'; the liquid product, whose yield was only 10-20% of that obtained when ethylene was used, contained 34.0% octanes but also 32.7y0 octenes. It has recently been shown that the alkylation takes place under milder conditions if a small amount (1-3% by weight of the total charge) of a halogen-, nitrogen-, or oxygen-containing compound is added to the mixture of paraffin and olefins (O'Kelly and Sachanen, 5). Suitable "homogeneous catalysts" included chloroform, benzyl chloride, bensal chloride, acetyl chloride, trichloroacetaldehyde, nitromethane, and tribromopropane. Both normal and branched chain parafibs were alkylated. Reaction occurred a t 300-400" a t pressures of about 3000 p.s.i. or higher. For example, the continuous alkylation of isobutane with ethylene at 427" and 2500-3000 p.s.i. pressure, in the presence of 1%by weight of a chlorinated naphtha, resulted in alkylate to the extent of 190% by weight of the ethylene charged. The hexane fraction (43% of the alkylate) consisted mostly of neohexane, which mas thus obtained in about 25% of the theoretical yield. Formation of neohexane appears to indicate that the free radical mechanism is again involved, the "catalysts" serving as chaininitiators. b. Catalytic Alkylation. The catalytic alkylation of isoparaffins takes place a t very much lower temperatures and pressures than does the thermal reaction. Yields are usually higher. Depending on the particular reactants and catalyst used, the temperature may be as low as -30" or as high as 100". The pressure is usually the vapor pressure of the reaction mixture and ranges from 1 to 15 atmospheres. Alkylation with ethylene takes place more rcadily in the presence of the metal halide catalysts than in the
ALKYLATION OF ISOPARAFFINS
31
presence of the acid catalysts. All of the catalysts are converted to addition complexes (so-called “lower layer” or “sludge”) with highly unsaturated compounds formed during the reaction. The products obtained by the condensation of isobutane with various olefins are not those to be expected from the simple addition of t-butyl and hydrogen across the olehic double bond. Alkylation with ethylene yields 2,3dimethylbutane and not 2,2-dimethylbutane1 which is the product of the thermal reaction. Similarly, alkylation with propene yields 2,3- and 2,kIimethylpentane rather than 2,2dimethylpentane. The principal alkylation products from 1-butene (especially in the presence of aluminum chloride modified by reaction with methanol) are dimethylhexanes whereas those from 2-butene are trimethylpentanes. By-product formation is due chiefly to two side reactions, namely “destructive alkylation” and a hydrogen-transfer reaction, the net result of which is the hydrogenation of alkene and the “self-condensation” of the isoalkane. CIHI,+CIH~~-~C~H~+C~HI~ 2C,Hio+CsHs +CsHla+CsH,
The significance of these side reactions is discussed below in greater detail. Catalytic alkylation of n-alkanes takes place only very slowly, particularly if conditions are not favorable for isomerisation ; side reactions often predominate. (1) Halide catalysts of the Friedel-Crafts type. The most important catalysts of this class are anhydrous aluminum chloride and boron fluoride promoted by small amounts of hydrogen chloride and hydrogen fluoride, respectively. It has been suggested that a function of the hydrogen halide is to convert the aluminum or boron halide to a more active acidic form. For example: :Cl: A~:CI: H :ci:%
+
:Cl:
[
1-
..:“‘:.* :ci: A ~ : C I : H+ :c1:
Recent work (Brown and Pearsall, 6) has indicated that the hydrogen aluminum tetrachloride is nonexistent, but that interaction of aluminum chloride and hydrogen chloride does occur in the presence of substances (such as benzene and, presumably, olefins) to which basic propert,ies may be ascribed. It has also been pointed out (Luder and Zuffanti, 7) that aluminum chloride and boron fluoride, being electron acceptors, are per se acids in the modern meaning of the word. Modified aluminum chloride catalysts (aluminum chloride sludges, aluminum chloride monomethanolate, aluminum chloride-nitroalkane
32
V. N. IPATIEFF AND LOUIS BCHMERLING
solutions, and aluminum chloride-alkali metal chlorides) have been investigated with the view of overcoming certain disadvantages of the unmodified catalyst. (2) Acid-type catalysts. Sulfuric acid of 96-98% concentration catalyzes the alkylation of isoparaffins with butenes and higher molecular weight olefins at 0-20'. Alkylation with propene requires acid of 98-10070 concentration and a temperature of 30". Alkylation with ethylene has not been very successful, presumably because stable ethyl hydrogen sulfate or diethyl sulfate forms too readily. When liquid hydrogen fluoride is used as catalyst, alkylation may be accomplished in satisfactory yield (except with ethylene) at room temperatures. Higher temperatures can be used with hydrogen fluoride than with sulfuric acid because it is only with the latter that a side reaction, oxidation of the alkene, must be avoided. . Both sulfuric acid and hydrogen fluoride have found wide commercial use aa catalysts for the alkylation of isobutane with propene and butenas. 11. THEMECHANISM OF THE CATALYTIC ALEYLATION OF PARAFFWB
As has already been mentioned, the catalytic alkylation of isoalkanes does not yield the anticipated isomers. Several mechanisms have been proposed to explain the formation of the products. The first of these (Xpatieff and Grosse, 8) postulated that the reaction proceeds via the addition of the l-alkyl group (corresponding to the isoalkane) and hydrogen to the olefin to yield a paraffin containing a quaternary carbon atom; this compound is not isolated but undergoes isomerization to form the isomeric alkane which is actually obtained. Thus, for example, the alkylation of isobutane with ethylene would yield 2,2dimethylbutane as the unstable intermediate and 2,3dimethylbutane aa the final product. A serious objection to this mechanism is that 2,2dimethylbutane is the predominant hexane in equilibrium mixtures of hexanes at temperatures up to 200' and even higher (Rossini, Prosen, and Pitzer, 9). Therefore, it is to be expected that if 2,2dimethylbutane were formed as the primary product, only a small part would rearrange and a substantial amount would be present in the alkylation product. A second mechanism (Rirch and Dunstan, 10) differs from the first in that it proposes that the alkylation involves a primary rather than the tertiary carbon atom of the isoalkane. Alkylation of isobutane with propene, for example, would occur by addition of isobutyi and hydrogen to the double bond to yield 2,4dimethylpentane. Isomerization of part of this heptane would then produce 2,3dimethylpentane which is also a major product of the reaction. It was postulated that the isoparaffin is activated
ALKYLATION OF ISOPARAFFINS
33
(via the formation of a complex with the catalyst, sulfuric acid) and dissociates into a negative alkyl group (in this example, isobutyl) and a proton, the proton coming from a methyl group. The alkene is activated by addition of a proton, furnished by the acid, to form a positive alkyl ion (isopropyl). Combination of the carbanion and the carbonium ion yields the paraffinic product (2,4-dimethylpentane). However, this mechanism has the same objectionable feature as the first; the proposed isomerization of the primary product is improbable. Two groups of investigators (Caesar and Francis, 11; hlcAllister et al., 12) have proposed mechanisms which are similar in that each assumes that the isoparaffin undergoes carbon to carbon cleavage to yield two alkyl fragments (isopropyl and methyl, in the case of isobutane) which then add to the olefinic double bond to yield a higher molecular weight paraffin. The mechanisms do account for the structures of the products obtained. However, to explain the formation of 2,3-dimethylbutane by the alkylation of isobutane with ethylene, it is necessary to assume that the ethylene acts as if it were ethylidene. A similar postulation for the alkylation of isobutane with propene (i.e., isopropylidene) would lead to the erroneous conclusion that 2,2,3-trimethylbutane is formed in substantial amount. A modified form of this mechanism has been proposed (Gorin, Kuhn, and Miles, 13). By assuming that the alkyl fragments (behaving as if positively and negatively charged) add to a complex formed between catalyst and olefin and by setting up rules governing the (‘preferred” form of the complex, it becomes possible for this type of mechanism to explain why some isomers are formed and others are not when isoparaffis are alkylated with olefins. However, this mechanim does not explain the formation of a11 of the products which are obtained when isoparaffins are condensed with chloroolefins (which may be considered to be “tagged” olefins) in the presence of aluminum chloride (Schmerling, 14b). Thus, for example, while it can explain the formation of l-chloro-3,4dimethylbutane by the reaction of isobutane with ally1 chloride, it cannot explain the formation of 1,2-dichloro-4,4dimethylpentaneas a by-product of this reaction. Furthermore, it does not account for the principal chlorinated product of the reaction of isobutane with vinyl chloride (namely, 1,ldichloro-3,3-dimethylbutane) or with 1,2-dichloroethylene (namely, 1,1 , s trichlor0-3~3-dimethylbutane). A mechanism which seems to give a truer picture of what occurs during alkylation (and which readily explains the formation of the polychloroalkanes as inherent products, see Section IV, lc) proposes that the reaction is a chain reaction involving the conversion of the isoparaffin into a t-alkyl ester (Schmerling, 14). The alkylation of isobutane with ethylene is used below in illustration, of this mechanism. Similar reactions occur with other
34
V. N. IPATIEFF AND LOUIS SCRMERLINQ
isoalkanes and alkenes. The separate occurrence of each: step was proved rigidly only for aluminum chloride as catalyst. However, since the alkylation products obtained with other catalysts, including hydrogen fluoride and sulfuric acid, are similar to those obtained with aluminum chloride, it may be presumed that analogous reactions occur with these other catalysts, the reaction intermediates being t-alkyl fluorides and t-alkyl hydrogen sulfates, respectively, rather than t-alkyl chloridcs. step 1
H
CIL -
A-I
c1
A
CHs+CHI = C H 2 + H C I - c ~ , - -CH;+CH, I
-CH,
The t-butyl chloride formed in step 3 starts a new cycle by reacting with ethylene as in step 2. Ethane is produced only in the initiating step and the amount formed is, therefore, small. Brief mention will be made here of only part of the considerable evidence which may be cited in support of each of the three steps. Thus, the first step (which probably occurs via the formation of ethyl chloride) finds confirmation in the fact that isoalkanes and alkyl halides undergo hydrogenhalogen exchange in the presence of aluminum chloride to yield t-alkyl halides and the alkanes corresponding to the original alkyl halides (Bartlett, Condon, and Schneider, 15). The primary reaction of talky1 chlorides with alkenes in the presence of aluminum chloride (i.e., step 2) is one of addition of the alkyl group and the halogen to the double bond of the alkene; a 75% yield of l-chloro-3,3dimethylbutaneis obtained by the condensation of t-butyl chloride with ethylene at -10' in the presence of aluminum chloride (Schmerling, 16). The third step involves a hydrogenhalogen exchange similar to that of step 1, except that it is accompanied by skeletal rearrangement. Direct proof that rearrangement of the carbon skeleton of the alkyl chloride occurs during its conversion to alkane is the fact that 2,3dimethylbutane and t-butyl chloride are major products of the reaction of l-chloro-3, 3dimethylbutane with isobutane in the presence of
ALKYLATION
35
OF ISOPARAFFING
aluminum chloride a t room temperature; comparatively little 2,2-dimethylbutane is formed (Schmerling, 14c). The rearrangement which occurs in step 3 may best be explained in terms of Whitmore’s theory of intramolecular rearrangements (Whitmore, 17); the reactions of steps 1and 2 may, of course, also be expressed in terms of “carbonium ions” (cf. Bartlett, Condon, and Schneider, 15). Thus, the alkylation of isobutane with 2-butene may be illustrated by the following equations (Schmerling, 14d) :
+
CH&H=CHCITs+H + +CH&€ICH&HS
+
+
CH~CHCH~+CII~CHCH~~H~~CH~CCH~+CH~CH&IIICH, CHI I
LHa
CHI
CHa
I
I
+
CH,C++CH&H=CHCHa-+CH,C-CHCHCH,
I
AH*
I
CRsCHS
(A)
CHs
I
+
CHIC-CHCHCH AHaCITa I (-4)
+
+
CHsC-CH-CHCHa+CHaCHCHs+CHsCH-CH-CHCHs+CHaCCHa I 1 I I I l l I CHsCHi CHa
CHs
CHa CHa CHa
CHI
(111)
The other octyl ions (I, 11, and IV) are similarly converted to the corresponding trimethylpentanes. Because relatively little 2,2,3-trimethyl*- Signifies D shift in position of the CHa group or the H atom. Read aa migration of the CHz group, or migration of the H atom.
38
V. N. IPATIEFF AND LOUIS SCRMERLING
pentane is obta&ed, it may be concluded that (I) is quite unstable and rearranges to (11) very readily. The alkylation of isobutane with 1-butene may be described by similar equations. The intermediate condensation product (A’) which corresponds to the 2-butene product (A) undergoes the following rearrangements:
-
II
-----t
CHs
I + I
-CHI
CfIaC-CHCHaCHzCH,--K!HaC
+-CHCH&HzC€i, I
1
CHI CHs
CHs
(V)
CHJ
+
I
,C&
CH,C-CH nCHCHi H-+C
I
+
HaCCHaCHCHZCH,
I
CHI
CHI AHl
(VI)
(Ail
CH, -H
I
+CHIC
+
-cH
+
-CHiCHaCIICHr--CH&CHsCH&HCH,
I
I
I
CHI
The tertiary carbonium ions (V), (VI) and (VII) are converted to dimethylhexanes by reaction with isobutane. The formation of (VII) may involve several intermediate hydrogen and methyl migrations rather than the 1,Cshift shown. Similarly] the fact that 3,4dimethylhexane is also a product of the alkylation indicates that the following changes occur:
+
+
hrH
CHICCH~CHCHZCH-+CH&HCHCHCH&HI
I
AH, AH,
CHI
(Vu
+
I
CH,
kCHa -Ha
+
CHICHiCCHCHaCHat-CHICHCH-CHCHnCHi
I
1
CHICHI
AH, AH,
Subsequent to the publication of the above discussed mechanim in condensed form, a partially similar mechanism waa proposed (Ciapetta, 18). The point of similarity is that the latter assumes that a tertiary alkyl carbonium ion (or, in the molecular sense, ester) adds to the olefin to yield a carbonium ion of higher molecular weight which then undergoes carbon-
ALKYLATION OF ISOPARAFFINS
37
skeleton rearrangement before being converted to the paraffin product. A very important point of difference is that Ciapetta does not use a chain mechanism for the continued conversion of the isoparaffin to the carbonium ion and of the olefin addition product to the final paraffin. Instead, he postulates that the isoparaffin lases a proton from the tertiary carbon atom to yield a carbanion which then loses a pair of electrons to form the carbonium ion. Tn the final step, the proton and the electrons add to the carbonium ion of higher molecular weight, forming the paraffin product. The chain reaction scheme involved in steps 2 and 3 shown above appears to be preferable because it offers a more logical and plausible explanation for the conversion of the isoparaffin to the reactive tertiary alkyl chloride or carbonium ion; it is very highly improbable that free electrons can exist in a reaction mixture in the manner proposed by Ciapetta. Furthermore, if both carbanions and carbonium ions are formed, some of their interaction product (2,2,3,3-tetramethylbutane in the case of isobutane) should be obtained. No evidence of the presence of this octane in an alkylation product has been found. Also, this mechanism does not account for the dichloroalkanes formed by the reaction of isobutane with ally1 or vinyl chloride. I t follows from the above discussion that the chain mechanism requires that markedly different products be obtained in the alkylation of isobutane with 1- and Z-butene, respectively. The former should yield dimethylhexanes and the latter, trimethylpentaneg, as the major products. Such has been shown (Schmerling, 14d) to be the case when aluminum chloride (particulary when modified to diminish side reactions) was employed as catalyst. The product of the alkylation with 1-butene in the presence of aluminum chloride monomethanolate contained about 60% by weight dimethylhexanes and 10% trimethylpentanes, whereas the 2-butene product contained 65% trimethylpentanes and only 4% dimethylhexanes. The difference in composition was apparent also in the A.S.T.M. octane numbers of the 125" end-point gasolines. That from l-butene had an octane number of only 76.1; that from 2-butene, 94.1. When, on the other hand, hydrogen fluoride is used as catalyst for the alkylation of isobutane, there is only a relatively small difference in the character of the products obtained with the two n-butylenes (Grosse and Linn, 19). The octane numbers of the 150" end-point gasolines prepared by the alkylation of isobutane with 1- and 2-butene, respectively, were 92.7 and 95.3. Furthermore, there was no significant difference in the composition of the respective products obtained by the alkylation of isobutane with s-butyl fluoride or with 2-butene in the presence of hydrogen fluoride (Linn, 20). Similarly, alkylation of isobutane with 1- and 2-butene, respectively,
38
V. N. IPATIEFF AND LOUIS SCHMERLINQ
in the presence of 97y0 sulfuric acid yielded gasolines (185" end point) of 92.9 and 93.0octane number (Birch et al., 21a). It has further been stated (McAllister et aZ., 12; cf. Birch et al., 21a) that "the reaction products obtained from the normal butylenes are quite similar to those obtained from isobutylene except for the amount, of light ends formed." Irreversible isomerization of l-butene into an equilibrium mixture of Zbutene and isobutylene was assumed to occur prior to reaction with the isobutane. (It is not necessary to assume isomerization to isobutylcne since the alkylates obtained with isobutylene and with Zbutene would consist largely of trimethylpentanes and, hence, would be quite similar). The difference between the behavior of the metal halide catalyst and that of the acid catalysts is readily explained (Schmerling, 14d) by postulating that the reaction which occurs most rapidly when a mixture of isobutane and 1- or 2-butene comes in contact with liquid hydrogen fluoride or sulfuric acid, is the addition of the acid to the olefin to form s-butyl ester. There then exists in the reaction mass an equilibrium mixture which may be described by the following: Eq.A. C=C -C -C+HX@C - C -C -C*C (Minor)
1
X
-(>C
(Major)
-C+HX
X = F or OSOaH
Octanes produced via the addition of t-butyl ester (from isobutane) to the butenes so formed will be the same regardless of what alkylating agent is used and the product will contain more trimethylpentanes than dimethylhcxanes. If all of the alkylate were produced through the olefins formed as in eq. (A), all three alkylating agents (l-butene, 2-butene and s-butyl ester) would yield identical products. However, part of the alkylate in the case of the butenes is formed by the direct reaction of the t-butyl ester with the olefin before the olefin has reacted with the acid catalyst; this accounts for the small differences in composition. With aluminum chloride catalysts, on the other hand, the reaction proceeds chiefly via the addition of t-butyl chloride to the original butene before more than a relatively minor portion of the latter can react with the hydrogen chloride promoter to yield s-butyl chloride. Therefore, the two n-butylenes yield very different alkylates. Experimental evidence in support of the postulation for the reactions that occur in sulfuric acid (or hydrogen fluoride) may be found in a study of the alkylation of isobutane with 2-butene in the presence of tritium sylfuric acid at 10" (Stewart and Denham, 2 2 ) . The results indicate that there was random distribution of hydrogen and tritium atoms in the olefin and the catalyst prior to alkylation. Also, when 2-butene was bubbled through tritium sulfuric acid at a rate such that there was little absorption,
ALKYLATION OF ISOPARAFFINS
39
hydrogen-tritium exchange occurred, demonstrating not only rapid exchange but also reversible absorption of the olefin in the acid. It was shown further that comparatively little exchange occurs when isobutane (and hence, alkylate) is treated with tritium sulfuric acid for a relatively long contact time (20 minutes). The relative difficulty in alkylating normal paraffins is presumably due to the fact that the exchange reactions of step 1 and step 3 do not take place as easily with hydrogen atoms attached to secondary carbon atoms as with those attached t o tertiary carbon atoms (cf. Bartlett, Condon, and Schneider, 15). Under the conditions required to obtain the hydrogenhalogen exchange reaction with normal paraffins, polymerization of the olefin (which may be considered to occur via a reaction similar to that of step 2) takes place very readily and becomes the predominant reaction. The probable function of hydrogen halides as promoters for the metal halides and boron fluoride in the alkylation of isoparaffins seems to be to initiate and maintain the formation of t-alkyl halide by the reaction of step 1. Also, it may convert the metal halide to the more active form, such as, for example, hydrogen aluminum tetrachloride or an ester thereof. 111.
SIDE REACTIONB
1. Destructive Alkylation I n virtually all cases, the alkylation of an isoparaffin with an olefin yields not only the products to be expected from the condensation of one molecule of the isoparaffin with one or more molecules of olefin but also paraffins of intermediate molecular weight. Thus, for example, pentanes, heptanes, and other alkanes containing an odd number of carbon atoms are obtained as by-products of the alkylation of isobutane with ethylene in the presence of aluminum chloride. Indeed, isopentane is usually formed in alkylation reactions involving isobutane regardless of the olefin or catalyst employed. The formation of these by-products is apparently due in part to further reaction of the primary products by dissociation into “new” paraffins and olefins and the subsequent reaction of these olefins with the original isoparaffin (as well as of the “new” paraffins with original olefin). Such reactions, which may be termed “destructive alkylation,” are similar to those which have been suggested to account for the products of the “autodestructive alkylation” of paraffins (Ipatieff and Grosse, 23). By-product formation in the alkylation of isobutane with propene, for example, may be indicated as follows:
40
V. N. IPATIEFF AND LOUIS SCHMERLINQ
CHI- CH -CH-CH* -CH,-+CHI-CH -CHg -CHI+ [CH* CHg] AH8 &IL (!!HI C b - CH -CHI+ [CHS- CHs]+CH1-CH -CH -CHI
HI
&HI
&€la
The over-all reaction may be written 2 C'H,,+CaHa~CsIIiz+CaHlc
It is apparent that the ratio of reacting isoparaffin t o olefin may thus be greater than unity and the yield of alkylate based on the olefin may be higher than theoretical. Such high yields have been obtained in the alkylation of isopentane in the presence of sulfuric acid or hydrogen fluoride (we below). In some cases, as in the alkylation of isobutane with ethylene, the destructive alkylation appears to involve the transfer of methylene. The dissociation may be considered to be either a dealkylation or a depolymerieation reaction. It presumably involves the intermediate formation of alkyl halide or other ester (or the corresponding carbonium ions) by hydrogen-halogen exchange followed by carbon-carbon cleavage. It has been pointed out (Bartlett, Condon, and Schneider, 15) that the criterion for ready cleavage of any carbonium ion is the possibility of forming a relatively stable smaller carbonium ion by cleavage between carbon atoms 2 and 3 (numbering from the carbon atom bearing the positive charge). Thus, during the alkylation of isobutane with propene,
+
CHI -C-CH
d&
-CHi
AHI
- CHa+CHa-C=C I
+
-CHa+CHs - CHI
CHI
The trimethyle thylethylene may undergo polymerization, copolymerization, or alkylation, or it may add a proton and then be converted to isopentane by means of the hydrogen exchange reaction. The ethyl carbonium ion loses a proton to yield ethylene which may then react with isobutane (via t-butyl carbonium ion) to yield 2,3-dimethylbutane. Insight into another mode of formation of these by-products is furnished by the observation that l-chloro-3,3-dimethylpentaneand l-chloro3,3-dimethylbutane are by-products of the condensation of ethylene with t-butyl chloride and t-amyl chloride, respectively, in the presence of aluminum chloride (Schmwling, 16). Partial conversion of either t-alkyl chloride into the other presumably occurs under the reaction conditions. It is apparent, then, that the alkyl chloride formed as indicated in step 3 of the three step chain mechanism may be converted into a chloride of higher or lower molecular weight before undergoing the condensation of
41
ALKYLATION OF ISOPARAFFINS
step 2, and that further reaction of the condensation product with isoparaffin yields the by-product paraffin and t-alkyl chloride in accordance with step 3. 2. Hydrogen Transfer Hydrogen transfer resulting in the conversion of more or less of the o l e h to paraffin always accompanies the catalytic alkylation of isoparaffins with oIefins. At the same time, some of the isoparaffin is converted to p a r a n containing twice the number of carbon atoms; thus, for example, trimethylpentanes are formed in all isobutane alkylations regardless of the olefin used. This side reaction may be explained (Schmerling, 14c, d) as follows in terms of the three step chain mechanism:
+ + + c-c-c+c-c-c+c-c-c+c-c-c C-C-C+H++X
+C-C-C+X-
I
C C f
I
+
L
L
c-c-c+c=c~-c-ic-c-c-c-c
c:
1:
C
C
+ I ' c-Lc-c-c+c-c-c+c-c-c-c-CSC-c-c
A 1C : +
I
C
L!C:
-Ii/-CHa
c-c-c-c-c+c-c-C4C-c-c-c-c+c-c-c
i:
C:L!
f
I
C
+ I C
The uver-all reaction may be written 2 C,H,o+CJls~C~H,,+CJH,
It is obvious that the production of trimethylpentane involves the formation of an equimolar quantity of a low-boiling paraffin corresponding to the olefin alkylating agent. Such was found to be at least semiquantitatively the case. It will be noted also that 2 moles of isobutane are consumed for each mole of liquid paraffin produced; this excessive consumption of the isobutane is usually undesirable in commercial alkylation processes. Direct evidence of this hydrogen transfer reaction in the presence of sulfuric acid was obtained in an investigation (Marschner and Carmody, 24) of the alkylation of isobutane with isoamylene (as t-amyl alcohol) and of
42
V. N . IPATJEFF AND LOUIS SCHMERLINQ
isopentane with n-butylene (as s-butyl alcohol) or with isobutylene (as t-butyl alcohol or t-butyl chluride). In each case, isoparaffin was formed in m o u n t equivalent to a large proportion of the olefin: isopentane was isolated in 40 to 50% of the theoretical yield based on the 2-amyl alcohol, and isobutane in 70-110~0yield based on the butyl derivative used. At the same time, much more than 1 mole of charged isoparaffin was consumed per mole of olefin derivative: about 1.8 moles of isobutane and about 2.2 to 2.7 moles of isopentane, respectively, reacted chiefly to yield octanes and decanes, part of which underwent destructive alkylation. The products of direct alkylation, nonanes, were formed in varying amount, depending on the reactants, the highest yield being obtained when s-butyl alcohol was used. When isobutane was allrylated with 1- or 2-butene in the presence of aluminum chloride monomethanolate, very little or no n-butane was formed despite the fact that appreciable amounts of 2,2,4-trimethylpentane were produced (Schmerling, 14d). Similarly, no n-butane or n-pentane, respectively, was obtained by the alkylation of isobutane with 2-butenc and with 2-pentene in the presence of sulfuric acid although trimethylpentanes were formed in both cases (McAllister et al.,12; cf. Marschner and Carmody, 24). This apparent discrepancy in the alkylation mechanism may be explained readily. The n-alkylcne is converted not into n-alkane, but into isoalkane. The resulting isobutane cannot, of course, be differentiated from that charged; the resulting isopentane, on the other hand, can be and was a-tually found in substantial yield. In other words, the proton transfer reaction is accompanied by rearrangement of the carbon skeleton of the carbonium ion. Not only the isoparaffin but also the unsaturated hydrocarbon tied up in the catalyst complex serves as a hydrogen donor in the transfer reaction. The latter thus becomes more highly unsaturated.
3. Polymerization Polymerization of the olefin is competitive with alkylation of the isoparaffin. It occurs under conditions which are unfavorable for rapid alkylation: low ratio of isoparaffin to olefin, spent catalyst, low catalyst to hydrocarbon ratio, too low temperature, and others. Polymerization of the olefin probably proceeds via the reaction analogous to step 2. The olefinic polymerization product may be isolated as such or it may undergo hydrogen transfer with the isoparaffin with the resultant saturation of the double bond. Also, hydrogen disproportionation between two molecules of the polymer may occur, yielding one molecule of paraffinic product and one of the ,highly unsaturated material found in the catalyst complex (cf. Bartlett, Condon, and Schneider, 15, p. 1536).
ALKYLATION OF ISOPARAFFINS
43
4. Formation of Catalyst Complex As has already been indicated hydrogen transfer reactions involving olefinic hydrocarbons lead to the formation of highly unsaturated compounds. These form addition complexes (“lower layer” or “sludge”) with the catalyst. In this way, crystalline aluminum chloride is converted to a red-brown liquid. Similar red to brown viscous products are obtained when sulfuric acid or hydrogen fluoride is employed. 6. Formation of Esters Alkyl esters are formed by addition of the hydrogen halide promoter used with aluminum halide or of hydrogen fluoride or sulfuric acid catalyst to the olefin. Their formation is also inherent in the reaction of the second step of the chain mechanism. If conditions are not favorable for the reaction of these esters with the isoparaffin (by the reaction of step 1or step 3), the esters are obtained as impurities in the alkylate. In general, they are produced under the same conditions which led to polymerization rather than alkylation. Usually very little ester is presnt in the alkylation product. IV. CATALYSTS 1. Aluminum Chloride a. With Olcfins. Although aluminum chloride was the first known catalyst for the alkylation of isoalkanes and has been intensively investigated, it has found less commercial application than the acid catalysts, suIfuric acid, and hydrogen fluoride. This is due primarily to the ease with which side reactions, particularly cracking, occur in its presence and to its relatively short life; special techniques must be employed in order to utilize it efficiently. On the other hand, in contrast with the acid catalysts, aluminum chloride readily catalyzes the alkylation of isoparaffins with ethylene in excellent yields and consequently has been used for the production of 2,3-dimethylbutane by the reaction of isobutane with ethylene. Alkylation with ethylene proceeds best at temperatures above room temperature and under superatmospheric pressure. Ethylation of isobutane in batch operation at 25-35” in the presence of aluminum chloride and hydrogen chloride resulted in a product that was 45% hexanes. The hexane fraction was shown to consist of 70-9070 2,3dimethylbutane, 1045% 2methylpentane, and less than 3y0 2,2-dimethylbutane (Grosse and Ipatieff, 25). The catalyst was converted into the typical red-brown liquid lower layer complex. Alkylation of n-hexane with ethylene takes place under similar conditions (Ipatieffet al., 26). It seems probable that isomerization to isohexane precedes the alkylation.
44
V.
N. IPATIEFF AND LOUIS SCHYERLING
Large scale ethylation of isobutane involves the usc of liquid aluminum chloride catalysts since these are more readily adaptable to continuous operation (Section IV, 2). When aluminum chloride is used at room temperature as catalyst for alkylation with olefins of higher molecular weight than ethylene, the reaction is accompanied by so much side reaction that the primary products no longer predominate. Formation of by-products may be greatly minimized by carrying out the reaction at a sufficiently low temperature, for example, -30 to -35" (Pines, Grosse, and Ipatieff, 27). Continuous alkylation of isobutane with propene at -30" and atmospheric pressure in the presence of unmodified aluminum chloride plus hydrogen chloride promoter yielded a liquid product that was 420/, heptanes (chiefly 2,3-dimethylpentane and some 2,4-dimethylpentane) and 20y0 decanes; 90% of the liquid boiled below 220". Alkylation of isobutane with n-butylene (33% l-butene, 67% 2-butene) under similar conditions (-35") produced a liquid product consisting of more than 60% octanes and 127; dodecanes. Analysis by means of the Raman spectra indicated that the octanes included 2,5-dimetliylhexane and 2,2,3-, 2,2,4-, and 2,3,4-trimethylpentane. Batch alkylation of isobutane with l-butene and with 2-butene at 30" resulted in deep-seated decomposition reactions, octanes comprising only 21-24% of the liquid products (Schmerling, 14d). Dimethylhexanes predominated in the l-butene product (6.4y0 trimethylpentanes, 11.5y0 dimethylhexanes, and 3.5% methylheptanes). Trimethylpentanes were the principal octanes present in the 2-butene alkylate (14.5% trimethylpentnnes, 5.8% dimethylhexanes, and 2.6y0 methylheptanes). The octane numbers of the 125" endpoint gasolines from the 1-butene and the 2-butene alkylations were 74.5 and 83.5, respectively. b. With AZkyZ Chlorides. Isopara&s (imlike aromatic hydrocarbons) undergo little alkylation when treated with alkyl chlorides in the presence of aluminum chloride; instead, intermolecular hydrogenation occur^, the alkyl halide is reduced to the corresponding paraffin and hydrogen chloride, the hydrogen being furnished by the original isoparaffi which is thereby converted either to the paraffinic "self-condensation" product or to the unsaturated hydrocarbon combined with the aluminum chloride in the lower layer. The over-all reaction is similar to that of hydrogen transfer with olefins (Section 111, 2). 2 i-C,H,,+GCjH,CI-.CsH,,
+Ca€Is +IIC1
The reduction is of course inherent in the hydrogen-chlorine exchange reaction which is the first step of the chain alkylation mechanism (Schmerling, 14). The reaction of isobutane with isopropyl chloride in the presence of
ALKYLATION OF ISOPARAFFINS
45
aluminum chloride a t 40 and 70" resulted in the formation of 60 and 90%, respectively, of the propane available by the reduction of the alkyl
chloride (Schmerling, 28). The isobutane was converted mainly to octanes (and products of their destructive alkylation) and to catalyst complex. No reaction at all occurred when isobutane was heated at 70" with aluminum chloride and hydrogen chloride. In an analogous manner, isopropyl chloride was converted to propane in 90% yield whenit was heatedat 40"with isopentane and aluminum chloride. Heating the isopentane with aluminum chloride and hydrogen chloride caused it t o undergo autodestructiva alkylation , yielding higher-boiling i s o p a r a h s and isobutane but no propane; less catalyst complex was formed than when isopropyl chloride was used. It has been claimed that isobutane is alkylated by methyl and ethyl bromide when aluminum bromide is used as catalyst. This point is discussed in Section IV, 6. c. With Chlmoolefins. Isobutane is chloroalkylated when it is contacted with allyl chloride in the presence of aluminum chloride at temperatures below about -10" (Schmerling, 14b). The major product, l-chloro3,4-dimethylpentane, is obtained in 3 5 4 0 % yield at -10"; 1,Zdichloro4,4-dimethylpentane is produced in 13-15%yield. On the other hand, what amounts to alkylation of the isobutane takes place when the reaction is carried out a t higher temperatures, the product consisting essentially of paraffinic hydrocarbons because of interaction of the chloroheptanes with the isobutane. At 0-5", for example, liquid paraffins were formed to the extent of 320% by weight of the propene available from the sllyl chloride; this represents the reaction of 1.6 moles isobutane per mole allyl chloride. The following mechanism, analogous to that suggested for alkylatiori with olefins, explains the experimental observations. At temperatures below about -10' : C Step 1'
Step 2'
step 3'
I
c -c -c+c = c -CCI-*C -c
-Cl+C
c:
C
C
C
!
=c-c
I
c -& - c1+c=c-CCI+C -c -c -c-c
c:
c:
I
Kb
A1 Al C
c - -c -c-c+c -c -c+c -c -c' c -CCl+C A1 &l
t:
t:
L-
-A
t:
-CI
46
V. N. IPATIEFF AND LOUIS SCHMERLINO
The l-butyl chloride formed in step 3' reacts with allyl chloride as in step 2', thus starting a new cycle. As previously mentioned, the reaction of the first step occurs only as an initiating step; the small amount of propene reacts either with isobutane (via t-butyl chloride) to yield heptane or with the catalyst to form lower layer complex. At temperatures above about -10':
c-c-c-c-cc1+c-c-c~c-c-c-c-c+cA cI
A
&A
i A
-c1
The l-Lutyl chloride may react with the catalyst t o form lower layer or may be converted into octene or octyl chloride which is then converted to octanes. The reletion of isobutane with vinyl chloride in the presence of aluminum chloride yielded liquid isoparaffiris and 1,l-dichloro-3,3-dimethylbutane (40y0yield a t -10' and 20% yield a t 25'). t-Rutyl was isolated in about 10% yield. The formation of the dichlorohexane instead of monochlorohexane may be explained as being due t o the relative unreactivity of the primary chlorine atoms; only a very small amount of the dichloride is reduced to monochloride by the reaction similar to step 3'. In the case of the allyl chloride reaction, the dichloroheptane contains a secondary chlorine atom and reacts readily with isobutane, yielding monochloroheptane and t-butyl chloride. Under conditions which are such ,as bring about the reduction of a chlorine atom of the dichlorohexane, both chlorine atoms are replaced, and the product consists of isopsraffins and only a very small amount of monochlorohexane. Similarly, the condensation of isobutane with 1,2-dichloroethylene yields 1,1,2-trichloro-3,3-dimethylbutaneas principal chlorine-containing product (Schmerling, 29). This trichloride is obtained in 35% yield by the reaction of isobutane with thc cis isomer of the dichloroethylene a t room temperature and in less than 5% yield by reaction with the trans isomer under the same conditions. 2. Aluminum Chloride Sludges
Because 2,3-dimethylbutane is a very desirable component of aviation gasoline and because aluminum chloride is a particularly suitable catalyst for the production of this hexane by the ethylation of isobutane, much work has been done to develop processes for carrying out the reaction on a commercial scale in continuous flow operation. Excellent results have been obtained with liquid aluminum chloride sludges consisting of aluminum chloride dissolved or dispersed in complexes of aluminum chloride with highry unsaturated hydrocarbons. These sludges are insoluble in the
ALKYLATION OF ISOPARAFFIXS
47
alkylate, settle rapidly and may be separated from it very readily and recycled. They may be prepared by a number of methods. They may be formed in situ by the alkylation reaction or be produced by the action of aluminum chloride on olefins or such substances as isooctane, kerosene, or similar hydrocarbon fractions. Two groups of investigators (Alden et al., 30 and Thompson and Chenicek, 31) have discussed the results obtained with the liquid sludge formed during the alkylation of isobutane with ethylene. Another group (Holloway and Bonnell, 32) has described the use of the sludge prepared by the action of aluminum chloride on the highboiling residue (b.p. 161-195”+) from the distillation of sulfuric acid alkylate (formed by the alkylation of isobutane with propene and butenes). An important advantage of the sludge catalysts is that they can be maintained a t a constant uniform level of activity by the addition of small amounts of fresh catalyst. When preformed sludge, such as that formed by the action of aluminum chloride on alkylate bottoms is used, the makeup catalyst may be pumped into the reaction mixture, an equivalent amount of spent catalyst being removed at the same time (Holloway and Bonnell, 32). When the sludge is formed in situ, fresh aluminum chloride may be added as such at spaced time intervals (Alden et al., 30) or continuously as a solution in the isoparafh. The latter is conveniently accomplished by passing all or a portion of the isobutane through a heated zone (a “saturator”) containing granular aluminum chloride (Thompson and Chenicek, 31). The bulk supply of the catalyst is not contaminated inasmuch as isobutane does not react with aluminum chloride at the temperatures used (about 60”) in the absence of a promoter. This method is flexible and exact since the amount added can be controlled by regulating the temperature or the rate of flow of the isobutane. It has the advantage that the separate sludge manufacture step is eliminated and that there is no need for relatively complicated fresh sludge pumping means. Furthermore, the yield of alkylate per unit weight of aluminum chloride is increased because no catalyst activity is wasted merely to form the sludge. Less side reactionoccurswith the sludge catalysts than with unmodified aluminum chloride. Excellent yields of alkylate containing as much as 80% by volume of hexane were obtained in the continuous alkylation of isobutane with ethylene (in about 5 to 1 ratio) at 60” in the presence of sludge plus fresh aluminum chloride added in isobutane solution using hydrogen chloride as promoter (Thompson and Chenicek, 31). Similar yields of product containing 70-7595 of hexanes were obtained at 56-59’ with preformed sludge and in the absence of added hydrogen chloride (Holloway and Bonnell, 32). In the former experiments, the reactor was a turbomixer; in the latter, a jet reactor was used. As has just been indicated, no hydrogen chloride was added in the
48
V. N. IPATIEFF AND LOUIS SCHMERLING
experiments made with the preformed sludge. Holloway and Bonnell state that since it was found possible to secure alkylation without the addition of hydrogen chloride and since this is advantageous, no data were accumulated on the effect of varying amounts of the promoter. They point out that the catalyst was prepared without the addition of promoter and that any free hydrogen chloride formed in its preparation was worked out of the sludge during the cooling and stirring procedure. Furthermore, free hydrogen chloride was not found in the catalyst upon analysis. They conclude, “Thus it does not appear that appreciable concentrations of free hydrogen chloride could have been present in these runs, although it cannot reasonably be denied that traces such as would be required in a recently proposed alkylation mechanism (Schmerling, 14c) may have existed.” Holloway and Bonnell are probably correct in their assumption thak only traces of hydrogen chloride (or other chlorine-containing compound which can act as chain initiator) need be present in order that alkylation can occur. It is furthermore reasonable to assume that an additional trace amount was continually added with the makeup sludge. On the other hand, experiments in which sludge formed in situ was used have shown that the presence of substantial quantities of promoter has a marked effect on the catalyst life and hence on the yield of alkylate per unit weight of aluminum chloride (Thompson and Chenicek, 31). It was found that when no promoter was used, the yield of alkylate, and the percentage of hexanes in the alkylate, decreased rapidly as the run progressed. Satisfactory results were obtained with from 0.5 to 5.2 mole % hydrogen chloride (based on the total charge). The composition of the products was excellent in all cases (the alkylate containing 79436% of hexanes) but the yield of alkylate decreased with increase in hydrogen chloride concentration from 2.1 to 5.2%, which suggested that ethylene was removed from the reaction by combining with the hydrogen chloride to form ethyl chloride. This was confirmedby analyzing the exit gas (ethyl chloride forms an azeotrope with n-butane) and it led to the use of ethyl chloride instead of hydrogen chloride as promoter. It was found that alkylate yields approaching the theoretical were obtained a t 60” in the presence of 1.8 mole yoethyl chloride; SO% by volume of the alkylate was hexane and 98% boiled below 150”. There was little net consumption of ethyl chloride. Traces of ethane were formed which indicates that the function of the ethyl chloride probably is to initiate the chain reaction of alkylation. The fact that ethyl chloride rather than hydrogen chloride is found in the existing gas may explain why Holloway and Bonnell found no hydrogen chloride in the sludge. Further evidence that addition of hydrogen chloride promoter has a beneficial effect may be seen in the fact that catalyst life is greatly increased.
ALKYLATION OF ISOPARAFFTNS
49
A life test lasting 54 days was carried out (Thompson and Chenicek, 31) ; during this time 65 g. of fresh aluminum chloride was introduced from the saturator and 47.1 1. of total alkylate was produced. This is equivalent to a yield of 85 gallons alkylate per pound aluminum chloride consumed. On the other hand, it was found (Holloway and Bonnell, 32) that the catalyst life obtained with sludge formed in situ (but using a jet reactor instead of a turbomixer) and in the absence of added promoter was 26 gallons alkylate per pound aluminum chloride. When the preformed sludge was used (again without added promoter) ,it was not found possible to obtain a catalyst life greater than 17.2 gallons alkyIate per pound aluminum chloride. The alkylate obtained in the life test reported by Thompson and Chenicek contained 80.7% by volume hexanes and about 12% octanes. The hexanes consisted of 92% 2,3dimethylbutane, 1-2y0 2,2dimethylbutane, and 7% of 2-methylpentane as shown by infrared analysis. Thus, more than 72% of the total alkylate was 2,3-dimethylbutane. The octane fraction was essentially 2,2,4-trimethylpentane. The effect of the composition of the aluminum chloride hydrocarbon complex on the yield and quality of alkylate was investigated by recycling the sludge at 60" without adding any fresh aluminum chloride; hydrogen chloride was used as promoter (Thompson and Chenicek, 31). It was found that the aluminum chloride content of the sludge slowly decreased because of the gradual accumulation of hydrocarbons in the complex. A catalyst of excessively high aluminum chloride content (90-9201,) gave a fair yield of alkylate (253-276% by weight of the ethylene charged) but its hzxane content wasonly 49-57% by volume. As the run progressed and the aluminum chloride content dropped to 82-84%, the alkylate yield was 300-3160/, and the hexane content was quite constant at 73-74%. The optimum aluminum chloride content appeared to be 63434%. That the increase in hydrocarbon content is not due to replacement of either chlorine or aluminum seems to be proved by the observation that the chlorine to aluminum atomic ratio was relatively constant at 2.2-2.5 throughout the course of a run. (The chlorine values were based on ionizable halogen. The fact that the ratio was not 3.0 is significant. It seems to indicate that chlorination of hydrocarbon material occurred with the resultant formation of hydrogen chloride or alkyl cgoride. Such chlorinecontaining material, much of which might be tied up in the catalyst complex, could react in the chain-initiating step of the alkylation, thus making addition of promoter unnecessary during the earlier part of the sludge's Use.) The effect of changes in a number of other variables (temperature, pressure, contact time, isobutme to ethylene ratio, and diluents) on the
50
V. N. IPATIEFF AND LOUIS SCHMERLING
yield and quality of the alkylate has also been investigated. Discussion of the results as well as of those obtained in commercial production of 2,3-dimethylbutane is, however, beyond the scope of this paper. 3. Aluminum Chloride Monomethanolate The addition complex of equimolar amounts of aluminum chloride and methanol is a crystalline compound which is stable at temperatures below 70"; a t higher temperatures it decomposes, forming hydrogen chloride and methoxyaluminum dichloride, further heating of which yields methyl chloride and oxyaluminum chloride (Norris and Sturgis, 33). I t very readily catalyzes the alkylation of benzene, in which it is soluble, and of isobutane, in which it is not (Schmerling, 14d). On the other hand, the product of the reaction of aluminum chloride with two molecular proportions of methanol (ie., AlCla-2 CHSOH) is not an alkylation catalyst. The activity of aluminum chloride monomethanolate promoted by hydrogen chloride is such that alkylation of isobutane with propene or butene takes place a t room temperature or higher with comparatively little side reaction, thus making possible the isolation of primary reaction products. Heptane comprised 52y0 of the liquid obtained by the reaction of isobutane with propene at 62". The heptane consisted of nearly equal amounts of 2,3- and 2,4-dimethylpentane. With l-butene as the alkylating agent a t 55", the alkylate contained 60% dimethylhexanes (35% 2,4- and 2,5-, 17% 2,3- and 8% 3,4-dimethylhexane) and 9.570 trimethylpentanes (6.5% 2,2,4- and 3% 2,3,4-trimethylpentane). With 2-butene at 28", the products included 65% trimethylpentanes (28% 2,2,4-, 22.5% 2,3,4-, 14% 2,3,3-, and 0.5% 2,ZJ3-trimethylpentane) and 4% dimethylhexanes (3.5% 2,4- and 2,5- and 0.570 2,3-dimethylhexane). With 2-butene at a higher temperature, namely 64", the trimethylpentanes comprised 44% and the dimethylhexanes 7YoJ of the liquid product. The octane numbers of the 125" end-point gasoline obtained with l-butenc and with 2-butene mere 76 and 92-94, respectively. Although the alkylation with l-butcne proceeded smoothly at 55", relatively little alkylation occurred at 28", a temperature which gave very satisfactory results wiFh Zbutene. That the ease of alkylation with 1butene a t the higher temperature mas not due to intermediate isomerization to 2-butene is evident from the composition of the products. It is worthwhile to compare these results with those obtained with pure aluminum chloride as catalyst a t 30" (Section IV, la). The significance of the fact that the major product of the alkylation with l-butene was dimethylhexanes while that with 2-butene was trimethylpentanes haa been discussed in the section on the mechanism of the reaction.
ALHYLATION O F ISOPARAFFINS
51
4. Aluminum Chloride-Nitroalkane Solutions Aluminum chloride dissolves in as little as its own weight of the readily available nitroparaffins, nitromethane, nitroethane, and the two nitropropanes. The resulting solutions, particularly those employing nitromethane, are catalysts for the alkylation of isobutane, hydrogen chloride serving as a promoter (Schmerling, 28). On the other hand, solutions of aluminum chloride in molecular excess of ethers, ketones, and alcohols are catalytically inactive, at least for the alkylation of hydrocarbons. The active component of the nitroparaffin solutions of aluminum chloride has been shown to be not the metal chloride per se but rather the addition compound, AlCla.RN02. In this connection it is significant to contrast the nitroparaffin solutions with the ether, ketone, and alcohol solutions. All four types of compounds form monomolecular addition complexes (usually crystalline compounds of low melting point) which are catalytically active (Schmerling, 34). For example, aluminum chloride monomethanolate, as was shown above, catalyzes the alkylation of isoparaffins. Solutions of the aluminum chloride monomethanolate in excess methanol are, however, catalytically inert; indeed, even the solid product of the reaction of aluminum chloride with two molecular proportions of methanol (Lea,A1C13.2 CHaOH) is not an alkylation catalyst. Similarly, aluminum chloride monoetherates and monoketonates lose their catalytic activity when dissolved in ethers and ketones, respectively. In direct contrast with these, the aluminum chloride nitroparaffin complexes retain their activity even with a large excess of the nitroparaffin solvent. a. With Olefins. Alkylation of isoparaffins with olefins took place in best yield when nitromethane was used as solvent for the aluminum chloride. The nitroethane and 1-nitropropane solutions gave fair to good yields of alkylated product while rather poor yields were obtained with the 2-nitropropane solutions. Higher reaction temperatures are required with these catalysts than with unmodified aluminum chloride. Alkylation of isobutane with propene at 35" in the presence of a nitromethane solution of aluminum chloride promoted by hydrogen chloride gave a yield of liquid product amounting to only 110% by weight of the propene charge; the yields of heptanes and of octanes were, respectively, 9 and 6% by weight of the propene. Carrying out the reaction at a higher temperature, 70°, resulted in a 200% yield of alkylate, a 105% yield of heptanes, and a 29y0 yield of octanes. The heptanes, which were thus produced in 44% of the theoretical amount, consisted chiefly of - ,3- and 2,4-dimethylpentane. The octane was formed by the hydrogen transfer side reaction ; propane was found in approximately equimolar quantity. The catalyst solutions were not soluble in the paraffinic product. They
52
V. N. IPATIEFF AND LOUIS SCHMERLINQ
underwent little change in weight during the reaction and were recovered as brown mobile liquids. b. With Alkyl Chlorides. It was shown in Section IV, l b that isopropyl chloride is converted to propane in 60-90% yield when it is treated with isobutane or isopentane in the presence of aluminum chloride. I t is, therefore, of interest to note that the use of nitromethane solutions of aluminum chloride permits the alkylation of the isoparaffins with isopropyl chloride. Less than 3oy0 of the isopropyl chloride was reduced to propane when it was heated with isobutane at 70" in the presence of the catalyst solution. The liquid product (looyob y weight of the available propene) consisted of a mixture of paraffins, heptanes, and octanes being formed in largest amount (35 and 20% respectively). The yield of liquid product was higher than that (70-75%) obtained in the presence of unmodified aluminum chloride. The reaction of isopentane with isopropyl chloride at 70" in the presence of the nitromethane solution of aluminum chloride yielded octane (42% by weight of the available propene) as the major product. Propane was formed in only 15% yield. Little or no reaction occurred when either isopentane plus hydrogen chloride or isopropyl chloride was heated at 6070" with the nitromethane solution of aluminum chloride. 6. Aluminum Chloride-Alkali Metal Chloride Double Salts
Aluminum chloride forms double salts of relatively low melting point (e.g., 140-300") with a large number of metallic chlorides. The use of such addition compounds with equimolar quantities of sodium chloride, lithium chloride, ammonium chloride, or potassium chloride as catalysts for the alkylation of isobutane with ethylene, propene, or isobutylene has been described (Blunck and Carmody, 35). Temperatures of at least 150" at pressures of about 1000 p.s.i. were found to be necessary for alkylation when a mixture of isobutane and the olefin waa passed, at a space velocity of 1-5 cc./cc. catalyst/hour, over the double salt deposited on pumice. The activity of the double salts decreased in the order mentioned, the potassium chloride compound being quite inactive but yielding some polymerization product at 316". The product obtained with ethylene in the presence of the sodium chloroaluminate catalyst a t about 150-170" was completely p a r a 5 i c but the yield was low: 75-90y0 weight of the ethylene charge. Higher yields were obtained with propane and isobutylene, but the product contained substantial quantities of olefins formed by the competing polymerization reaction. Catalyst life was very poor. However, it should be noted that hydro-
ALKYLATION OF IBOPARILFFMS
53
gen chloride promoter was not used in this study. Further investigation is required before final conclusions as to the catalytic behavior of these double salts can be made.
6. Aluminum Iiromide No detailed discussion of the use of aluminum bromide as catalyst for the alkylation of paraffins with olefins has been published. Alkylation presumably takes place in much the same manner as with aluminum chloride, the chief difference being that the catalyst is soluble in the paraffin. Alkylation of n-butane and isobutane with methyl or ethyl bromidein the presence of aluminum biomide has been claimed (Heldman, 36). I t was found, for example, that the action of aluminum bromide (0.00158 mole as AlzBrs) on a solution of 0.0392 mole methyl bromide in 0.0750 mole n-butane a t 25" for 120 hours yielded 0.0052 mole isopentane (13% of the theoretical based on the methyl bromide) ;isobutane was also formed. I n a similar experiment carried out a t 78" for 65 hours, the yield of isopentane was 33%. The reaction of isobutane (0.0520 mole) with methyl bromide (0.0392 mole) in the presence of aluminum bromide (0.00142 mole AlzBrs) a t 25" for 283 hours yielded low boiling material (0.0009 mole), n-butane (0.0062 mole), unreacted isobutane (0.0410, 79% of that charged), isopentane (0.0034 mole, 9% of the theory) and higher boiling material (0.0024 mole). Analogous results were obtained with ethyl bromide. The action of aluminum bromide on a solution of 0.0262 mole ethyl bromide in 0.0500 mole n-butane at 25" for 42 hours yielded 0.0150 mole material boiling higher than n-butane. With isobutane (0.0741 mole), ethyl bromide (0.0642 mole), and aluminurn bromide (0.00310 mole Al2Bre) a t 50" for 48 hours, there was obtained a trace of methane, 0.0140 mole ethane (22% of the theoretical based on the ethyl bromide), 0.0080 mole propane, 0.0082 mole n-butane, 0.0185 mole (25% of that charged) isobutane, 0.0130 mole isopentane (20 mole yo of the ethyl bromide charge), and 0.0095 mole hexane and higher. The present authors question whether these results prove that the alkyl bromides (particularly methyl bromide) actually underwent condensation with the butanes. I n most of the experiments no data are given t o show how much methane or ethane was formed. For none of the experiments is the amount of unreacted methyl or ethyl bromide recorded. It is quite possible that hydrogen-halogen exchange occurred and that the isopentane and the higher-boiling paraffins were formed entirely from the butane.
54
V. N . IPATIEFF AND LOUIS SCHMERLINC)
CII, -CHI -CHs -CHs+CHsBr+CHa-CH
-+
- CHn -CHa+Bf
+CHI
+ 1T + CHa-C -CHaF),CHa-CH -CH* I
AH8
CHa
+
Also, CHa-CH-CHt+CHaBr-+CHa-C-CHa+BrAH8
+CHI
LHa
The t-butyl carbonium ion (or t-butyl bromide) may undergo disproportionation reactions of the type outlined in the sections on destructive alkylation and hydrogen transfer to yield isopentane and higher-boiling paraffins. Further experiments are, therefore, required before it can be stated definitely that the methyl group of the methyl bromide is present in the final alkylate. The desired information might be obtained by carrying out larger scale tests with more accurate analysis of all of the products. It would almost certainly be obt,ained from experiments employing methyl bromide containing radioactive carbon. If mcthylation of the butane predominates, the isopentane will be radioactive; if reduction of the type shown above occurs, the radioactive carbon will be found in the methane.
7. Boron Fluoride Boron fluoride, promoted by a minor amount of hydrogen fluoride or water, seems to catalyze the alkylation of isoparaffins with ethylene a t lower temperatures than does aluminum chloride (Grosse and Ipatieff, 25). The reaction of isobutane with ethylene to -30 to -40" and 6 atmospheres pressure yielded a liquid product, 20% of which was hexanes, to the extent of 180% by weight of the ethylene charged. At 0-5" and 10 atmospheres pressure, a 224% yield of alkylate, 45% of which was hexanes, was obtained. Like the aluminum chloride product formed at 25-35', the hexanes consisted of 70-90% of 2,3-dimethylbutaneJ 10-25% of 2-methylpentane, and 3% of 2,2-dimethylbutaneS Alkylation of isobutane with isobutylene at 25" in the presence of boron fluoride promoted by water yielded an alkyIate which contained 32% octanes and 15% dodecanes (Ipatieff and Grosse, 37). 8. Zirconium Chloride The action of Eirconium chloride in the alkylation of isoparaffins with olefins and in other hydrocarbon reactions is very similar to that of rtluminum chloride, the principal difference being tliat it requires higher temperatures. Comparatively little information concerning its use has, however, been published.
ALKYLATION OF ISOPARAFFINS
55
Alkylation of isobutane with ethylene in the presence of zirconium chloride took place a t 100" under 15 atmospheres pressure (Ipatieff, 1, p. 682). The product was completely paraffinic and consisted chiefly of hexanes, octanes, and decanes. The catalyst was converted to a dark, pasty mass which was still catalytically active as was shown by its reuse in a second experiment with isobutane and ethylene. The alkylation of isobutane with n-butylene was subsequently carried out by gradually adding an equimolar mixture of the parsffin and the olefin to a stirred mixture of zirconium chloride.and a minor amount of isobutane in the presence of hydrogen chloride (Ipatieff , Friedman, and Schmerling, 38). At 23" and 6 atmospheres pressure, there was formed a liquid product (bromine number, 18) which contained 81% material boiling below 225", of which about 467, was octanes and 187, was dodecanes. The liquid product obtained a t 50" and 7 atmospheres had a lower bromine number,3; it contained 79% of material boiling below 225", of which about 587, was octanes and 19% was dodecancs. The octanes consisted principally of trimethylpentanes. The A.S.T.M. octane number of the material boiling a t 95-116' (85% of the total octane fraction) was 91.
9. Sulfuric Acid a. With Olefins. Concentrated sulfuric acid is an excellent catalyst for the alkylation of isoparaffins with propene and higher molecular weight olefins; it does not readily catalyze alkylation with ethylene, presumably because stable ethyl hydrogen sulfate is formed. Best results are obtained with 9 6 9 8 % acid. Stronger acid is required for alkylation with propene than with the higher-boiling olefins; acid strengths as high as 101.77, have been used with propene. The acid forms complexes (sludges) with highly unsaturated hydrocarbons during the alkylation and gradually loses its activity as an alkylation catalyst. Water formed by side reactions during the alkylation or introduced with the hydrocarbons furt,her serves to dilute and weaken the catalyst. In the continuous alkylation of isobutane with propene and the butylenes the acid concentration can drop to 90% without markedly affecting the product; with amylenes, 85% acid is satisfactory. If the acid concentration drops below these strengths, polymerization of the olefin becomes the chief reaction. In commercial operation the concentration of the acid is maintained at the desired level by withdrawing a portion of the used acid and adding makeup catalyst of 96-100% strength. Suitable alkylation temperatures lie in the range 0-30". The optimal temperature depends on the olefin and the catalyst strength. In general, higher temperatures can be employed with propene than with the butylenes and higher-boiling olefins. The lower the temperature, the higher is the strength of the acid which is required. Thus, alkylation of isobutane with
56
V. N. I P A m F F AND LOUIS SCHMERLING
propene in the preeence of 9874, sulfuric acid proceeded smoothly at 30", yielding 213% alkylate by weight of the propene; 90% of the alkylate was aviation gasoline boiling below 150" (McAllister et al., 12). Little or no alkylation occurred in the presence of 970/, sulfuric acid at 20"; instead isopropyl hydrogen sulfate and polymers and hydropolymers were obtained (Birch et al., 21b). Satisfactory results were obtained a t this temperature by employing a stronger acid, 101.7%; 216% by weight of alkylate, over 90% of which boiled below 150" was obtained. Theoctane numbers of the aviation gasoline fraction obtained in the two experiments cited were virtually the same, 88.5 and 89.1, respectively. When the butylenes or amylenes are used aa alkylating agents, temperatures above 20" result in poor yields of product based on the weight of both the olefin and the catalyst because oxidation of olefin occum with the accompanying formation of sulfur dioxide. At 20" with 97y0 sulfuric acid, alkylation of isobutane with l-butene and 2-butene1 respectively, resulted in 159 and 164% yields of alkylate, containing 88 and 90% 185" end-point gasoline of 89.1 and 90.2 octane number when a 1 to 1ratio of isoparafih to olefin was used (Birch et al., 21a). With a 4 to 1 ratio, the respective alkylate yields were 183 and 199%; the 185" end-point gasolines (92 and 9l%, respectively, of the alkylates) had octane numbers of 92.9 and 93.0. At 10" with 1 0 0 ~ sulfuric o acid, a 2000/, yield of alkylate, 93% of which boiled below 150"and had an octane number of 94, wasobtained (McAllister et al., 12) by the alkylation of isobutane wit,h 2-butene (6 to 1 ratio). When isobutylene was the alkylating agent (7to 1 ratio), use of 98% sulfuric acid at 20" yielded 180y0 of alkylate, only 81% of which boiled below 150" and had an octane number of 91.5 (McAllister et al., 12). In the commercial alkylation of isobutane with the butylenes, temperatures between 0 and 10" are employed because catalyst life is increased with a decrease in temperature. The strength of the makeup acid is 98100%. Temperatures below 0" are not used because the emulsified mixture of catalyst and hydrocarbons becomes too stiff to flow readily. Intimate contact of the catalyst and the hydrocarbon mixture is essential for high yields of alkylate and low acid consumption. Mole ratios of isoparafin to olefin in the range 5 to 1-10 to 1 and volume ratios of acid to hydrocarbon of about 1 to 1 give the best results. Contact times of about 20-40 minutes are usually employed. Typical yields of alkylate obtained in plants operated during World War I1 were in the order of 1-2 gallons per pound of catalyst in the alkylation of isobutane with butylenes and only about 0.5 gallon per pound for alkylation with propene or amylenes (Anonymous, 39). The octane numbers of the alkylates (175" end point) obtained with butylenes, propene, and amylenes were about 92-94, 88-89, and 89- 90, respectively.
ALKYLATION O F ISOPARAF'FINS
57
Alkylation of isopentane with 2-butene at loo in the presence of 100% sulfuric acid resulted in a 264% yieId of alkylate based on the olefin (McAllister et al., 12). The theoretical yield for the simple alkylation reaction is 228y0. Isobutane, hexanes (chiefly 2-methylpentane), nonanes, and decanes were formed. Destructive alkylation and hydrogen transfer occurred (Section 111). Neohexane did not react when contacted with 2-butene a t 10" in the presence of 100% sulfuric acid (McAllister el al., 12). Under similar conditions, alkylation of isohexane (a mixture of 2- and 3-methylpentane) yielded alkylate to the extent of 206% by weight of the 2-butene. It therefore appears that isoparaffins that contain tertiary carbon atoms are more susceptibIe to aIkylation. The composition of a large number of alkylates, both those prepared in the laboratory using pure hydrocarbon8 and those obtained on a commercial scale using olefin mixtures, has been investigated. The principal ' products are for the most part similar to those obtained in the presence of the metal halide catalyst, the chief difference being in the relative amounts of the various isomers. 'Alkylation of isobutane with propene a t 30" in the presence of 98y0 sulfuric acid yielded an alkylate that was 6246% by weight 2,3dimethylpentane and 8-12%, 2,4dimethylpentane; propane and trimethylpentanes (ll-19y0) were also obtained (McAllister et al., 12). A commercial isobutane-propene alkylate obtained a t 21" w~tsshown to contain about 36% by volume of 2,3- and 26% of 2,4dimethylpentane as well as 15% of trimethylpentanes (Glasgow et aE., 40). The product obtained by the alkylation of isobutane with 2-butene a t 10" in the presence of 100% sulfuric acid contained 34-38% 2,2,4-trimethylpentane and 51-55% 2,3,4- and 2,3,3-trimethylpentane (McAllister el al., 12). n-Butane was not obtained; the significance of this observation was discussed above (p. 42). Alkylation of isobutane with isobutylene a t 20" in the presence of 98% acid yielded an alkylate which contained 24-28% by weight of 2,2,4-trimethylpentane and 30-3401, of 2,3,4- and 2,3,3-trimethylpentane, as well as 7-9% of isopentane and 8-10% of 2,3dimethylbutane formed by destructive alkylation (McAllister el al., 12). The significance of the fact that alkylation of isobutane with l-butene yields a product which is very similar to that obtained with 2-butene has already been discussed (Section 11). Because of the importance of the alkylate prepared from isobutane and butylene in the aviation gasoline program during World War 11, a number of commercial products were analyzed, with the view of determining the factors which would result in the maximum yields of fuels of high octane number (Glasgow et al., 40; Field and Gould, 41). The A.S.T.M. octane numbers of three alkylates produced under decreasing throughput
58
V. N. IPATIEFF AND LOUIS SCHMERLINQ
ranged from 91.9 to 96.7, the best product being obtained by use of an extremely high ratio (33.5) of isobutane to butylene. The analytical results showed that all three alkylntes contained essentially the same relative proportions of the four trimethylpentanes but the total trimethylpentane contcnt of the alkylate varied from 56.1% by volume for the product of lowest octane number to 86.7% for that of highest octane number. Less dimethylhexane was prescnt in the latter nlkylate. Evidence of a rather large extent of hydrogen transfer may be found in experiments with isobutane and olefins (particularly those containing tertiary carbon atoms?) of higher molecular weight than the butylenes (McAllister et al., 12; cf. Marschner and Carmody, 24, discussed in Section 111, 2). Thus, the allcylation of isobutane with 2-pentene yielded a product containing 6-870 of isopentane, 14-22Y0 of trimethylpentanes, and 5ti-65Y0 of nonanes; that obtained with 2-methyl-2-butene contained 18-20% of isopentane, 29-33% of trimethylpentanes, and only 1 5 2 0 % of the expected nonanes. Similarly, alkylation of isobutane with the octene prepared by t h e dehydration of 2-ethyl-l-hexanol resulted in a product containing 35% of 3-methylheptane. Alkylation of isobutane with propylene trimers yielded nonane (45oj, of the alkylate) which corresponded closely to that obtained by hydrogenating the trimer; trimethylpentanes (33-400/, of the alkylate) were formed in approximately the same molar amount as the nonanes. Hydrogen transfer was also the principal reaction that occurred when isobutane was contacted with diisoamylene a t 10' in the presence of 100% sulfuric acid. Decanes and trimethylpentanes comprised 16-20% and 40-50a/,, respectively, of the alkylate; isopentane (10-12%) was also produced, presumably by depolymerisation of the diisoamylene followed by hydrogen transfer. No nonanes (formed by depolyalkylation) or tetradecanes were obtained. Depolyalkylation does occur when isobutylene polymers are used (McAllister e l al., 12). Thus the reaction of isobutane with butylene trimer at 10" produced an alkylste containing 6 0 4 5 % trimethylpentane and 10-15y0 dodecanes which were presumably formed by hydrogen transfer; the relative yields of trimethylpentanes and dodecanes indicate that depolyslkylation occurred. A similar product was obtained by the reaction of isobutane with hot acid codimer of isobutylene and n-butylene (largely 2,2,4-, 2,2,3-, and 2,3,4-trimethylpentenes). These observations were confirmed in commercial alkylations. Alkylation with the dimer of 2-butene yielded an alkylate which had quite similar boiling point properties but a lower octane number (88 as compared to 93 for each of the other products). The presence of dimethylhexanes, formed by hydrogen transfer, is indicated. b. Wilh AZkyZ Sulfates. The alkylation of isoparaffins with alkyl hydrogen sulfate or with dialkyl sulfate has received little study. Only one
ALKYLATION OF ISOPARAFFINS
59
experiment appears to have been described (Birch et al., 21b). The reaction of isobutane with diiosopropyl sulfate at -12' in the presence of commercial sulfuric acid resulted in a 65% yield (by weight of the propene "content" of the diisopropyl sulfate) of hydrocarbon with a boiling range of 88-244". No reaction occurred when the sulfate and isobutane were contacted under similar conditions but in the absence of sulfuric acid. The use of higher reaction temperatures for the acid catalyzed reaction is indicated. Determination of the composition of the product would be of interest in connection with the alkylation mechanism. c. With AZcohols. Alkylation of isopnraffins with alcohols was carried out in an investigation (Marschner and Carmody, 24) of hydrogen transfer (cf. p. 41). The reaction of isobutane with t-amyl alcohol at 2" in the presence of 98.6y0 sulfuric acid yielded 248y0 liquid paraffins (including isopentane formed by hydrogen transfer) by weight of the amylene available from the alcohol; the theoretical yield of alkylate is 183y0. The product contained about 20% by weight isopentane, 46% octanes, and 9% nonanes. The nonanes represented 13y0 of the theoretical yield; the octanes and isopentane, 70 and 5oyOyields, respectively, assuming ,the interaction of two moles of isobutane with one of alcohol. Similarly, alkylation of isopentane with t-butyl akohol under the same conditions resulted in the formation of alkylate (boiling above pentane) to the extent of 3Q5%by weight of the available isobutylene. The product contained about 21% by weight hexanes, 6% heptanes, 9% octanes, 27% nonanes, and 25% decanes. Nonanes were thus formed in 36% of the theoretical yield. Decanes and isobutane, respectively, were isolated in 3oyO and 111% of the yields obtainable if all of the t-butyl alcohol had undergone hydrogen transfer and none simple alkylation to form nonanes. Extensive destructive alkylation of the decanes apparently occurred. Alkylation of isopentane with s-butyl alcohol at 24" resulted in 44% of the theoretical yield of nonanes and 18 and 72% yield decanes and isobutane, respectively, as well as substantial amounts of hexanes and octanes. d . With AZkyZ Chlorides. t-Butyl chloride has also been used as an alkylating agent for isoparaffins in the presence of sulfuric acid (Marschner and Carmody, 24). The reaction of isopentane with t-butyl chloride at 13' in the presence of 98.6% sulfuric acid resulted in a 262% by weight yield (based on the isobutylene available from the chloride) of hexanes and higher-boiling paraffins. Isobutane, nonanes, and decanes were formed, respectively, in 87, 27, and 28% of the theoretical yields. 10. Hydrogen Fluoride a. With Olefins. Like sulfuric acid, hydrogen fluoride readily catalyzes
the condensation of isoparaffins with propene and higher molecular weight
60
V. N. IPATIEFF AND LOUIS SCHMERLING
olefins (Linn and Grosse, 19). Unlike sulfuric acid, it is very stable and is not readily oxidized or reduced. This chemical property, together with its low melting and boiling points, -83" and 19. 4 O , respectively, gives hydrogen fluoride a number of advantages as an alkylation catalyst. The reaction can be carried out over a wide range of temperatures and the catalyst may be recovered and reused almost indefinitely. In commercial alkylation , for example, atmospheric or slightly superatmospheric temperatures are used, making refrigeration unnecessary; in the laboratory, temperatures as low as -25' or even lower may be employed. The spent catalyst can be regenerated merely by heating. Not only is the uncombined hydrogen fluoride thus distilled from the catalyst sludge, but the sludge itself is decomposed and the hydrogen fluoride which it contains is recovered. Similarly, substantially all of the fluorine content of the alkylate is recoverable as hydrogen fluoride by thermal or catalytic treatment. Catalyst consumption is, therefore, very low. Sixty to eighty-five gallons of alkylate per pound of hydrogen fluoride are obtained in commercial operation (Anonymous, 891. Anhydrous hydragen fluoride is the preferred catalyst because it does not corrode iron equipment under the alkylation conditions and because the presence of more thaa a small percentage of water has a detrimental effect on the yield of alkylate. Thus, hydrogen fluoride containing 10% by weight of water yielded isopropyl fluoride but practically no alkylate when employed as catalyst for the alkylation of isobutane with propene at 25" (Linn and Grosse, 19). Under the same conditions, use of a catalyst containing 1% of water resulted in a 214% yield of alkylate by weight of the propene. Somewhat more water can be tolerated with higher molecular weight olefins. The reaction of isobutane with n-butylenes at 25" in the presence of catalysts containing 1% and 10% of water, respectively, gave 199 and 192% yields of alkylate; hydrogen fluoride containing 26% of water produced s-butyl fluoride and almost no alkylate. Activity of the catalyst is also decreased by dilution with nonaqueous substances such as hydrocarbons in the catalyst sludge (Linn and Grosse, 19). The effect is smaller than that of dilution with water and the catalyst remains active until the hydrogen fluoride concentration drops below about 80%. In typical commercial operation, fresh (recycled) catalyst contains 85-90y0 titratable acid, about 1.5% water, the remainder being the highly unsaturated hydrocarbons in the catalyst complex. As has already been mentioned, alkylation in the presence of hydrogen fluoride may be carried out over a wide range of temperature. In the alkylation of isobutane with isobutylene, for example, increasing the reaction temperature from -24" to +32' had little effect on either the yield (198 and 195%, respectively) or the quality (91.5 and 92.0 A.S.T.M.
ALKYLATION O F ISOPARAFFINS
61
octane number, respectively) of the product (Linn and Grosse, 19). Further increase in the temperature to 60' caused a decrease in the yield of alkylate (177%) and an increase in the amount of destructive alkylation; the octane number of the alkylate was 89.0. Other variables may also be changed over a rather wide range withput any marked effects. Suitable conditions, as exemplified by those used in commercial plants, include volume ratios of catalyst to hydrocarbon ranging from about 0.5 to about 1.5, and mole ratios of isoparaffin to olefin ranging from about 5 to about 10. The contact time may be as low as onehalf minute or as high as several hours; contact times of about 15 to 30 minutes are used in commercial operation. Branched chain paraffins undergo isomerization and decomposition on long contact with hydrogen fluoride (Linn, 42; cf. Webb and Gallaway, 43). The alkylations are carried out under sufficient pressure (usually about 100-150 p.s.i.) to keep the major portion of the hydrocarbon in the liquid phase. The composition of the alkylates obtained by the reaction of isobutane with pure olefins at relatively low temperatures indicates that it is possible to obtain products which consist largely of the primary products, that is, those of molecular weight equal to the sum of the molecular weights of the isoparafh and the olefin. Such "selective alkylation" of isobutane occurred in batch experiments carried out at 10" using an isoparaffin to olefin mole ratio of about 3 and a catalyst to hydrocarbon volume ratio of about 0.15 (Linn and Grosse, 19). The alkylate obtained with propene contained 50% heptane consisting largely of 2,3dimethylpentane together with a smaller amount of 2,4dimethylpentane; about 10% 2,2,4-trimethylpentane was also present. Alkylation with l-butene, 2-butene, and isobutylene, respectively, resulted in alkylates, the distillation curves of which were very similar. About 80% of the alkylate consisted of octanes. In the case of the two n-butylenes, this was roughly half 2,2,4-trimethylpentane; the remainder consisted of the octanes boiling at 110-1 14", 2,3,4-trimethylpentane being the largest single component. In the case of isobutylene, about two-thirds of the octane fraction wm 2,2,4-trimethylpentane. The aviation gasoline cuts (150" end point) of the l-butene, Zbutene, and isobutylene products were 92.7, 95.3, and 96.7, respectively. Explanation of the fact that the n-butylene products were so similar has already been given (Section 11). Unpublished results (Linn, 42) indicate that still more selective alkylation of isobutane occurs at lower temperature, -20 to -30". Alkylation of isopentane with propene at 10" resulted in a 272% yield of liquid alkylate by weight of the olefin; 19% of isobutane mas also formed (Linn and Grosse, 19). The yield of alkylate (the theoretical is 272%) and
62
V. N. IPATIEFF AND LOUIS SCHMERLINQ
the fact that only 26% of the alkylate boiled in the octane range indicate that hydrogen transfer and destructive alkylation reactions occurred. As in the cage of the sulfuric acid products, a large number of commercial hydrogen fluoride alkylates were investigated during World War I1 in order to correlate yields, octane numbers, and compositions of the aviation gaaolines produced (Glasgow et al., 40; Field and Gould, 41; Webb and Gallaway, 43). A detailed discussion of the results and their significance can not be presented here. I t will suffice to indicate that it was found that the hydrogen fluoride alkylates were quite similar t o the sulfuric acid products except for a higher content of 2,2,4-trimethylpentane. An alkylate obtained by the reaction of isobutane with propene in the presence of hydrogen fluoride contained 16.5y0 by volume 2,2,4-trimethylpentane and 41y0 heptanes and had an A.S.T.M. octane number of 87.5 (Glasgow et al., 40, Field and Gould, 41); another product contained 19% 2,2,4-trimethylpentane and 56y0 of heptane and had an octane number of 89.7 (Webb and Gallaway, 43). An analogous sulfuric acid alkylate contained 8% 2,2,4trimethylpentane and 65.5% heptanes and had an octane number of 88.7 (Glasgow et al., 40; Field and Gould, 41). I t may be concluded that more side reaction, namely hydrogen transfer, occurred in the presence of hydrogen fluoride. Similar differences were observed with alkylates prepared by the reaction of isobutane with butylenes (Glasgow e l al., 40; Field and Gould, 41). Thus, a hydrogen fluoride nlkylate having anA.S.T.M. octane number of 91.1 contained 41.7, 2.9, 9.4, and 10.O~o(total, 64.0y0) 2,2,4-, 2,2,3-, 2,3,4-, and 2,3,3-trimethylpentane, respectively. An analogous sulfuric acid product contained 26.8, 1.6, 19.2, and 14.9% (total, 62.57,) of the respective trimethylpentanes; its octane number mas 92.3. b. With Alkyl Fluorides. Alkyl fluorides also serve as alkylating agents for isoparaffins in the presence of hydrogen fluoride (Linn, 20). The products are similar to those obtained when olefins are used; they contain only traces (0.01 to 0.1%) of fluorine. Batch alkylation of isobutane with isopropyl fluoride a t 37" resulted in a 226% yield of alkylate by weight of the propene available from the alkyl fluoride; hcptanes and octanes were present to the extent of 41 and 29y0, respectively, by weight of the alkylate. Alkylation of isopen tane with isopropyl fluoride under the same conditions resulted in a 247% yield of alkylate containing 23% octanes and 42% decanes and higher boiling product; hydrogen transfer and destructive alkylation occurred to a large extent. Alkylation of isobutane with s-butyl fluoride produced a 206% yield (by weight of the butene available from the alkyl fluoride) of alkylate, the composition of which did not differ significantly from that of the alkylate
ALKYLATION OF ISOPARAFFTNB
obtained when 2-butene was used as alkylating agent. implications of this observation are discussed on p. 38.
63 The theoretical
REFERENCES 1. Ipatieff, V. N., Catalytic Reactions at High Pressures and Temperatures. Macmillan, New York, 1936, pp. 673-701. 2. Egloff, G., and Hulla, G., Chem. Rev. 37, 323 (1945). 3. Frey, F. E., and Hepp, H. J., Ind. Eng. C h m . 28, 1439 (1936). 4. Oberfell, G. G., and Frey, F. E., Refiner Natural Gasoline Mfr. 18, 486 (1939). 5. O’Kelly, A. A., and Sachanen, A. N., Ind. Eng. Chem. 38,463 (1946). 8. Brown, H. C., and Pearsall, H., Paper presented at the New York meeting of the American Chemical Society, September, 1947. 7. Luder, W. F. and Zuffanti, S., Electronic Theory of Acids and Bases. Wiley, New York, 1946, Chap. 10. 8. Ipatieff, V. N., and Grosse, A. V., J . A m . Chcm. SOC. 67, 1616 (1935). 9. Rossini, F. D., Prosen, E. J. R., and Pitzer, I<. S., J . Research Nall. Bur. Standards 27, 528 (1941). 10. Birch, S. F., and Dunstan, A. E., Trans. Faraday Soe. 86,1013 (1939). 11. Caesar, P. D., and Francis, A. W., Znd. Eng. Chem. 33,1426 (1941). 12. McAllister, S. H., Anderson, J., Ballard, S. A,, and Ross, W. E., J . Org. Chem. 6, 647( 1941). 13. Gorin, M. H., Kuhn, C. S., Jr., and Miles, C. B., Ind. Eng. Chem. 38, 795 (1946). 14. Schmerling, L., (a) J . A m . Chem. SOC.66, 1422 (1944); (b) ibid. 67, 1438 (1945); (c) ibid. 67, 1778 (1945); (d) ibid. 68, 275 (1946). 15. Bartlett, P. D., Condon, F. E., and Schneider, A,, J . A m . Chem. SOC.66,1531 (1944). 16. Schmerling, L., J . A m . Chem. SOC.67, 1152 (1945). 17. Whitmore, F. C., J . A m . Chem. SOC.64,3274 (1932). 18. Ciapetta, F. G., Paper presented at the Nen York meeting of the American Chemical Society, September, 1944. Znd. Eng. Chem. 37,1210 (1945). 19. Linn, C. B., and Grosse, A. V., Znd. Eng. Chem. 37,924 (1945). 20. Linn, C. B., Paper presented a t the New York meeting of the American Chemical Society, September, 1944. 21. Birch, S. F., Drinstnn, A. E., Fidlcr, F. A., Pim, F. B., and Tait, T., (a) Znd. Eng. Chem. 31, 884 (1939); (b) ibid. 31, 1079 (1939). 22. Stewart, T. D., and Denham, H., J . A m . Chern. SOC.68, 1135 (1946). 23. Ipatieff, V. N., and Grosse, A. V., Ind. Eng. Cliem. 28,461 (1936). 24. Marschner, R. F., and Carmody, D. R., Paper presented a t the Atlantic City meeting of the American Chemical Society, April, 1946. 25. Grosse, A. V., and Ipatieff, V. N., J . Org. Chem. 8,438 (1943). 26. Ipatieff, V. N., Grosse, A. V., Pines, H., and Komarewsky, V. I., J . A m . Chem. Soc. 68, 913 (1936). 27. Pines, H., Grossc, A., V. and Ipatieff, V. N., J . A m . Chem. Soe. 64,33 (1942). 28. Schmerling, L., Paper presented tlt the New York meeting of the American Chemical Society, September, 1947. 29. Schmerling, L., J . A m . Chem. SOC.70,379 (1948). 30. Alden, R. C., Frey, F. E., Hepp, H. J., and McReynolds, L. A., OilGas. J . 44, No. 40, 70 (1946). 31. Thompson, R. B., and Chenicek, J. A., I d .Eng. Chsr 40, 1265 (1948).
.
64 32. 33. 34. 35. 36. 37. 38.
V. N. IPATIEFF AND LOUIS SCEMERLING Holloway, C., Jr., and Bonnell, W. S., Znd. Eng. Chem. 38, 1231 (1946). Norrk, J. F., and Sturgis, B. M., J. A m . Chern. Soc. 61, 1413 (1939).
Schmerling, L., Unpublished results (Universal Oil Products Company). Blunck, F. H., and Carmody, D. R., Znd. Eng. Chem. 32,328 (1940). Heldman, J. D., J. Am. Chem. SOC.66,1791 (1944). Ipatieff, V. N., end Grosse, A. V., J. Am. Chem. Soc. 67, 1616 (1935). Ipatieff, V. N., Friedman, B. S., and Schmerling, L., Unpublished rcsults (Universal Oil Products Company). 39. Anonymous, Petroleum Refiner 26, 180, 280, 354 (1947). 40. Glasgow, A. R.. Jr., Streiff, A. J., Willingham, C. B., Rossini, F. D., Pr, c . Am. Petroleurn Znst. III, 26, 137 (1946). Also J . Research Natl. Bur. Standards 38, 537
(1947). 41. Field, H. W., and Could, D. W., Proc. A m . Petroleum Inst. 26, 170 (1946). 42. Linn, C. B., Unpublished rcsults (Universal Oil Products Compuny). 43. Webb, G. M., and Gallaway, W. S,, Petroleum Processing 2,442 (1947).
m,
Surface Area Measurements A New Tool for Studying Contact Catalysts* P. H. EMMETT Mellon Institute, Pittsburgh, Pa.
CONTENTS ’I. Introduction . . . . . . . . . . . . . . 11. Measurement of Surface Areas by Low Temperature Adsorption Isotherms 1. History of Method . . . . . . . . . . . . 2. “Point B” Selection . . . . . . . . . . . . 3. Derivation of the Brunauer, Emmett, Teller (B.E.T.) Equation . . 4. Some Advantages of the B.E.T. Equation . . . . . . . 6. The B.E.T. Equation for Narrow Capillaries . . . . . . 6. Other Methods for Measuring Surface Areas by Gaseous Adsorption . . . 111. Limitations and General Critique of the Gas Adsorption Methods 1. Molecular Size . . . . . . . . . . . . . 2. The k value of Harkins and Jura . . . . . , . . . a. Modifications and Other Derivations of the B.E.T. Equation . . . 4. Confirmation of the B.E.T. Method on Nonporous Solids . . . 6 . Confirmation on LargePore Solide. . . . . . . . . 6. Area Measurements on Charcoal and Other Fine-Pore Solids . . . 7. Limitations and Precautions in Using the B.E.T. Method . . . . IV. Application of Surface Area Methods in the Study of Solid Catalysts . 1. Activity vs. Surface Area . . . . . . . . . . 2. Measuring Surface Promot.er Distribution . . . . . . . 3. Area Measurements and Chemisorption . . . . . . . 4. Application to Fischer-Tropsch Catalysts . . . . . . . 5. B.E.T. Equation and Pore Size Meesurements . . . . . . 6. Sintering Studies , . . . . . . . . . . 7. Present Status of the B.E.T. Method . . . . . . . . Referencee . . . . . . . . . . . . . . . ~
Page 65 66 66 67 68 72 73 75 77 77 78 79 79 80 80 81 81 81
82 85 85 86 88 88 89
I. INTRODUCTION Recently on reviewing the various five-year periods of development in cata ysis between the two World Wars, Taylor (1) stated: “The final fiveyear period (1935-1940) of our survey gave to contact catalytic science a tool of which it had long been in sore need.” He was referring to ti new method that had been worked out for measuring the surface areas of catalysts or other finely divided or porous solids by means of the physical ’
Contribution from Gulf Research & Development Company Fellowship, Mellon Institute, Pittsburgh. Pa.
65
66
P. H. EMMETT
adsorption of gases close to their boiling points. In describing the experimental procedure involved, he continued : “The technique is a standardized one which can be reproduced by different workers in different laboratories with considerable accuracy and reproducibility. It, therefore, becomes a norm for expressing numerically what many believe to be very close to the accessible surface of a solid body.’’ He went on to express the hope that such surface area determinations would become a requircd part of all catalyst publications and catalyst patent applications. In view of the general importance that seems likely to be attached to this new tool for measuring catalyst surface areas, it seems worthwhile to restate occasionally its details and to discuss critically new developments in its theory and application in catalytic work. Accordingly, in the present chapter there is first presented a review of the gas adsorption method as originally developed and published, and then a critical discussion of recent papers dealing with its new modifications, derivations, and applications to catalysts and catalytic materials.
BY
11. MEABUREMENT OF SURFACE AREAS Low TEMPERATURE ADSORPTION IBOTHERMS
1. History of Method The general idea of measuring the surface areas of catalysts by adsorption methods is not new. The earliest attempts, however, made use of chemisorption rather than physical adsorption. For example, Benton (2) attempted to measure the surface area of platinum catalysts by the chemical or activated adsorption of CO and I&; de Boer and Dippel (3) suggested measuring the surface area of CaFa crystals by the chemisorption of water vapor; and Emmett and Brunauer (44 found it possible to estimate the iron surface of a catalyst by the low temperature chemisorption of CO a t - 195’ C. However, all methods employing chemisorption are subject to the grave limitation of finding a gas which can be shown to be chemisorbed by the entire surface rather than by a fraction of the surface of the catalyst without, however, reacting with more than a single outer layer of catalyst atoms or molecules. Physical adsorption has long been known and in a general way has been recognized as a means of obtaining some idea as to the relative surface areas of solid materials. For example, Russell and Taylor ( 5 ) compared promoted and unpromoted nickel catalysts by noting the relative amounts of COa adsorbed by them at -78” C. However, physical adsorption, like chemical adsorption, to be effective in measuring surface areas must be used in such a way as to enable one to be sure that he can determine the amount of gas required to form acomplete monolayer on thesolid adsorbent.
BURFACE AREA MEASUREMENTS
67
Only within the last 15 years has serious attention been directed toward ascertaining ways and means of selecting on physical adsorption isotherms the point corresponding to the amount of gas required to form a monolayer on catalysts or other adsorbents. About 15 years ago, it became increasingly evident that a method was needed to measure the surface areas of iron synthetic ammonia catalysts in an effort to differentiate between promoter effects which merely involved the maintenance of a large surface area and those which increased the activity per unit amount of area. The method had to be applicable t o active catalysts without destroying or altering their activity. It occurred to the author that a suggestion in a paper by Benton and White (6) might furnish a clue to a possible procedure for making such measurements. On obtaining an S-shaped adsorption isotherm for nitrogen on an iron catalyst a t -191.5’ C.they had pointed out that the rather sharp break at about 120 mm. pressure might mark the completion of a first layer and the beginning of a second layer of adsorbed gas. Clearly, if the point on such an isotherm corresponding to the completion of a monolayer of adsorbed gas could be found, a simple multiplication of the number of molecules in the layer by the cross-sectional area of the molecules being used as adsorbate would yield an absolute value for the surface area. Accordingly, the adsorption isotherms for a number of gases in theneighborhood of their boiling points were measured on an iron synthetic ammonia catalyst. The first group of results obtained (Emmett and Brunauer, 4a) is shown in Fig. 1. These isotherms marked the launching of a research program that culminated in a simple and effective method for measuring the surface area of catalysts or other finely divided or porous materials. %. “Point B” Selection A cursory examination of the isotherms in Fig. 1showed that the sharp break found by Benton and White was absent. The disappearance of this break resulted from applying perfect gas corrections to the gas phase of the adsorbate gas in the adsorption bulb at liquid air temperatures. However, the corrected isotherms still seemed to contain a point that might well be related to the completion of a monolayer. All of them were characterized by a long linear portion extending usually over several hundred millimeters pressure. Furthermore, gases having molecules of approximately the same size all showed linear portions beginning a t approximately the same volume of adsorbed gas. It seemed reasonable, therefore, to conclude that for the gases tried, the long linear part of the isotherm might represent the building up of a second layer of adsorbed gas. On this basis, the beginning of the long linear part of the isotherms of the type shown in Fig. 1 would represent the point of completion of a monolayer; for convenience, it was labeled
68
P. H. EMMETT
“point B.” A series of adsorption isotherms for nitrogen, argon, carbon dioxide, carbon monoxide, and oxygen at low temperatures on iron and also on a variety of other finely divided or porous solids yielded a number of types of evidence recounted elsewhere (Emmett and Brunauer, 4a and b;
FIG.1. Adsorption itjotherma on iron catalyst (No. 973) for various gases near their boiling poiuta (44. Curve 1A is for the total CO adsorption at -183°C. Curve IB represents the adsorption obtained rtt - 183” C.after evacuating at - 78” C. the aample for which the isotherm 1A had been obtained. Solid symbols are for desorption.
Emmett, 28, 33) that added to the conviction that the “point B ” method could be used for these gases to measure the surface areas of contact catalysts provided the measurements were made a t temperatures close to the boiling points of the respective adsorbates. However, similar adsorption experiments with butane soon showed that for many isotherms the “point B” method was inapplicable. Such a curve for butane is shown in Fig. 2, together with isotherms of other gases. Clearly it would be nearly impossible to select a “point B” on an isotherm that is so nearly a straight line through the origin as is the butane curve. 9. Derivation of the Brunauer, Emmett, Teller (7) (B.E.T.)Equution
In view of the apparent failure of the “point B” method to cover all adsorbates and adsorbents, an attempt waa made to develop a theory for the multilayer adsorption that might enable one to select the point on
SURFACE AREA MEASUREMENTS
69
adsorption isotherms corresponding to a monolayer even for gases which yield isotherms of such a shape as not to be amenable to the “point B” method. This resulted in the derivation of an equation known as the Brunauer, Emmett,, Teller (7) equation that can be used to evaluate V,,
I
I
I
P, mm
FIG.2. Low temperature adsorption isotherms on 0.606 g. silica gel ( 4 4 .
the volume of gas in an adsorbed monolayer on the adsorbent. This equation has proved to be widely applicable in measuring the surfaces of catalysts by a variety of adsorbed gases and will now be briefly described. Originally the B.E.T. equation was derived by a kinetic approach similar to that used by Langmuir (8) in developing his monolayer a d s o rp tion equation. Indeed, Langmuir (9) also envisioned the possibility of
70
P. H. EMMETT
multilayer adsorption but did not actually carry through the derivation of a multilayer equation. By letting SO,s1, 92, etc., represent the values for the absolute surface area covered by 0, 1, 2, etc., layers of adsorbed molecules respectively, a series of equations was obtained (Brunauer, Emmett, and Tcller, 7) of the form alpso = b,sle - E I / R l '
and aipisi-1
= bisie-lCi/Rl'
(4)
--
where El, E2, E3, Ei are thc heats of adsorption in the indicated multilayers; all bl, az, ba, etc. are constants; and p is the pressure of the gaseous adsorbate. The surface area of the adsorbent is then given by the equation
CSi m
A=
(5)
0
and the total volume of gas adsorbed by the equation
where Vo is the volume of gas required to cover 1 sq. cm. of the surface with a unimolecular layer. It then follows that m
v
v
Avo
Vm
-=-=-
Cisi
o
C m
si
0
where V , M defined above, is the volume of gas required to form a monolayer over the entire catalyst surface. The summation in eq. (7)can be carried out as indicated in the original paper if one makes the simplifying assumptions that Ea= E3= E; and that the ratio of the coefficients for desorpt,ion and adsorption in the multilayers, bz/aa, b3/a8etc. are equal t o each other. The summation finally leads to the equation
--.
V
m
(C-1)2 =l/vmc+--
VmC
where z is the relative pressure, p / p o , at which a particular volume of gas V (expressed as cc. S.T.P.) is adsorbed by thc catalyst. Clearly a plot of the left side of the equation against x permits the slope and intercept to be used
SURFACE AREA MEASUREMENTS
71
for the evaluation of the constant C and of V , . The constant C is really albl E1-EL equal to - e RT where E l - EL is the difference between the average &bi
heat of adsorption in the first layer and the heat of liquefaction of the adsorbate. A few typical examples of the use of the equation are shown in Fig. 3 in which the isotherms from Fig. 2 are replotted, according to the
5.0
4.0
n
0 3.0 U
-
-
6
a
-no I
P/ Po
FIG.3. B.E.T.plots (7)for the adsorption of various gases on 0.606 g. silica gel (cq. 8). Curve 1, carbon dioxide a t -78"; 2,argon at -183"; 3,nitrogen a t -183"; 4, oxygen a t - 183"; 5, carbon monoxide at - 183"; 6, nitrogen st -1195.8"; and 7, butane (C4H10)a t 0°C.
multilayer B.E.T. equation. Linear plots are obtained over the relative pressure range extending from about 0.05 to 0.35 when eq. (8) is applied to the adsorption of nitrogen on the synthetic ammonia catalysts.
P. H. EMMETT
72
4 . S0.m Advantages of the B.E.T. Equation The first test to which eq. (8) was subjected was a comparison between V, values for nitrogen adsorption isotherms obtained by the equation, and those estimated by the “point B” method. Table I shows the results of such a comparison (Brunauer, Emmett, and Teller, 7). It is obvious that for nitrogen isotherms the agreement between the two methods of obtaining V , is good. In fact, it is probably well within the error involved in selecting “point B” on an experimental isotherm. TABLE I Values of Constants for Adsorption of Nitrogen at .W.l”K. on Various Adsorbenls Subdance
Unpromoted Fe catalyst 973 I Ilnpromotcd Fc catalyst 973 I1 Fe-AlpOccatalyst 954 Fc-AlaO, catalyst 424 Fe-A120rKI0 catalyst 931 Fc-AlpOa-KZO CfLtalyRt 958 Fe-K,O catalyst 930 Fused Cu catalyst Commercial Cu catalyst Cr,O, gel CrlOl glowcd Silica gel
V,
oo./g.
0.13 0.29
Point B cc./g.
2.86 2.23 0.81
0.12 0.27 2 78 2 09 0.76
.56
.65
.14
.12 .05 .lo 50.5 6.14 127.0
.05 .09 53.3 6.09 116.2
Application of eq. (8) to the adsorption isotherms for butane on silica gel showed one of the several ways in which the B.E.T. equation is superior to and more useful than the (‘point 13” concept. As illustrated in Fig.2 the adsorption isotherm for butane on silica gel is of such a shape as to make the picking of a point corresponding to a monolayer by the “point B” method very difficult if not impossible. Yet, the data when plotted according to eq. (8) yield a value for the surface area of the gel of 383 sq. meters/g. compared to a value of 477 sq. meters/g. obtained by nitrogen at - 195’ C. One further advantage of the B.E.T. equation over the (‘point B” method should perhaps be pointed out in passing. In Fig. 3 it can be seen that the intercept for the plots for nitrogen adsorption are quite small. Accordingly, only a small change in the slope of the linear plot would be required to make the curve pass through the origin. Hence, the possibility exists of taking only a single adsorption point in the range 0.2-0.3 relative pressure and connecting this point on a H.E.T. plot with the origin. The slope of this line should, for an itdsorbent for which the nitrogen adsorption
-
73
SURFACE AREA MEASUREMENTS
isotherm is the normal S-shaped variety, yield a surface area value that will, as a rule, be too small by only about 1-5%. This clearly means a big advantage over the “point B” method since the latter necessitates the obtaining of sufficient adsorption points to demarcate clearly the linear part of the isotherm.
kr.( €0.9)
44(
B.ET (EO. 10 CARBON 4 n = I . 15 Vm ~ 4 2 5 . 6 C = 79.6
I
Ic
-2 360 E
v,
0
B.L.1.
R€T
cj
a
CARBON 2 n 8 1.30 V m = 318.3 C 144
W
a n W m 280 n
a W
5
3
J
9 B.ET (€0.9) RE.T. (EO. 10)
200 CARBON 5 n 11.45 Vm 179.I C a224
120
t a6 0.2
0.8
0.4
I
P/ Po
Fro. 4. Admxption isotherms for nitrogen at - 195’ C. on three samplesofcharcoal. The curves corresponding to the “n” form of the B.E.T. equation (eq. 9) and to the Langmuh form (eq. 10) are shown for comparison. The n, C , and V,values a8 well as the data are taken from the paper of Joyner, Weinberger, and Montgomery (10).
6. The B.E.T. Equation jor Narrow Capillaries
In deriving eq. (81,the authors (7)assumed that the summation indicated in eq. (7) cou!d be carried out to infinity. Clearly, however, there
74
P. H. EMMEW
might be many reasons for having to limit the summation to n layers. For example, if a capillary with plane parallel walls 2n molecular diameters apart is exposed to a gas at a temperature at which multilayer adsorption could occur, no more than n layers on an average could be built up on each v,* 100 c 100 9
€0. 9 V -I
a
EO.
u
10
2 .o
8
( n e w )
I
I
3.0
4.0
n
FIG.5. Error involved in using cq. (10) and (8) as approximations for eq. (0) at various values of n. The figure IVW taken from the paper by Joyncr, Wcinberger, and Montgomery (10).
face of the capillary. To take such a case into account, the original fundamental assumptions were modified by summing to "n" layers rather than to an infinite number of layers. There was then obtained the equation V=
VmCZ(1 -(n+l)zn +nzn+') (1 -2) (1 (C - 1)z - czn + I )
+
where all of the symbols are the same as in eq. (8) and n is the maximum number of layers of adsorbate that could be built up on the wall of a capillary without meeting a similar layer from the opposite wall. One interesting property of eq. (9) comes to light if one lets n = 1. It can be shown (7) that with such an assumption the cquation reduces to the form p / V = polCVm+plVm.
(10)
This is identical to one of the forms in which the Langmuir equation can be written except that the empirical Langmuir constants are replaced by the more readily definable constants V,, po and C. Equation (9) is much more difficult to apply than is eq. (8). However,
SURFACE AREA MEASUREMENTS
75
Joyner, Weinberger, and Montgomery (10) have shown a method by which, without too much labor, one can evaluate C , Vm, and n for materials in which the capillaries are so narrow as to force one to use eq. (9) rather than eq. (8). The details of their method need not be discussed here since they are we!l explained in their original paper. However, in Fig. 4 are shown several curves for the adsorption of N Pon charcoal together with the values for n, C , and V , obtained b y them (Joyner, Weinberger, and Montgomery, lo). They not only outline the method for applying eq. (9) but point out the percentage difference in the value of V , that is obtained by using either of the two extreme eq. (8) or (10) for isotherms for which n is really intermediate between 1 and 5. Figure 5 is their plot of the error involved in using the simplified equations as a function of the value of n. For example, if n is really 1.5 but is assumed to be 1 in order to use eq. (lo), the calculated surface area will be too large by about 15y0. On the other hand, if for such an adsorbent the area is evaluated b y use of eq. (8),it will be too s m d by about ISYO. With the help of eq. (9) and the procedure they outlined it becomes possible to evaluate the surface area even of catalysts having pores so small as to render impracticable the use of eq. (8).
6. Other Methodsfor Measuring Surjdte Areas by Gaseous Adsorption a. Procedure of Askey and Feachem ( 1 1 ) . Several other methods for measuring surface areas with the help of gaseous adsorption isotherms have been published. Askey and Feachem (11) pointed out in a n article published in 1938 that for several years they had been using low temperature adsorption isotherms for measuring the surface areas of finely divided materials. Their method is substantially the same as the first suggestion of Brunauer and Emmett (12) according to which it was postulated that an extrapolation of the linear part of the S-shaped adsorption isotherm to zero pressure should yield a value for the volume of gas required to form a monolayer. A simple multiplication of the number of molecules in this monolayer by the average cross-sectional molecular area would then yield a n absolute surface area. This method was later judged by Emmett and Brunauer (4a) t o be less accurate and more open to question than the “point €3” method; it certainly is less well founded than the method (7) using eq. (8). On the other hand, it must be admitted that this zero pressure extrapolation method certainly will yield areas which are of the correct order of magnitude and which are probably capable of yielding fairly reliable relative areas for various finely divided solids. b. Work of Harkins and Jura (13). A distinctly different approach to the interpretation of low temperature gas adsorption isothcrrns has been followed by Harkins and Jura (13). By assuming that thc same type of
76
P. H. EMMETT
equation is applicable to the adsorption of gases on solids that has been found applicable for correlating the surface spreading force of adsorbed films on liquids with the area occupied per adsorbed molecule, they have been able to obtain the equation log p = B - A / V *
(11)
where V is the volume of gas adsorbed at pressure p and A and 3 are constants. Furthermore, they deduce that the area of a solid can be obtained from a plot of log p vs. l / V z by means of the equation Area = k ( S ) X
(12)
where S is the slope of the plot of eq. (11) and k is a constant that had to be evaluated by some independent means. In some very ingenious experiments they (Harkins and Jura, 14, 15) also provided an independent means for evaluating k. By exposing a finely divided solid such as TiOz to a sufficient pressure of water vapor to form four or five statistical adsorbed layers and then immersing this water vapor coated sample in liquid water, they were able to obtain from the evolved heat a direct measure of the surface area of the powder without any assumption as to the molecular cross-section of the water molecule. They merely divided the heat evolved on immersing the sample by 118.5 ergs, the value for the normal surface energy per square centimeter of liquid water, to obtain a value for the number of square centimeters of area in the sample. Using this surface area for the powder they were then able to evaluate the constant k of eq. (12). For other solids they could apply eq. (12) directly by assuming that the constant k was independent of the type of surface. Proceeding in this way, they obtained the surface area of six separate solids from their nitrogen adsorption isotherms and compared them with the areas that were obtained by plotting the adsorption data according to eq. (8). The results are shown in Table 11. Comparison of the values in column 2 with those in column 6 of Table I1 shows that the agreement between the two independent methods of interpreting the nitrogen isotherms is clearlyremarkable and servesto strengthen the general conviction that gaseous adsorption isotherms can be used to obtain reliable surface area values for catalysts and other finely divided materials. Several papers (Livingston, 16; Emmett, 17) have been written to try to explain the queer coincidence that such divergent interpretations of the low temperature adsorption isotherms should lead to the same surface area values for the adsorbents. The agreement is still very puzzling and cannot be said to have been satisfactorily explained. However, attention has been called to the fact (17) that apparently the good agreement indicated by the nitrogen data in Table I1 is not always obtained. As a matter of fact it seems to be true that only when the C value of eq. (8) is about 100 will the
77
SURFACE AREA MEASUREMENTS
area values obtained by eq. (8) and by eq. (11) and (12) of Harkins and Jura (13) agree. Nevertheless, since many gases (especially those boiling near - 190" C.) yield C values in the range between 50 and 150, the agreement between the two methods is for the most part very satisfactory. Indeed, it is not yet possible to tell which method gives the more reliable area for those materials in which a divergence (which incidentslly never exceeds 30 or 40%) between the two methods exists. TABLE I1 Areas,, o j Solids Calculated by the Neil) .Ifelhod of Harkins and Jtira arid by that of Brunarier, Emmett and Tellerb New method of Harkins and Jurn
Solids
I
Nq
Hz0
TiO, (Standard) 13.8 13.8 TiO, I1 8.7 8.4 SiO, (Quartz) 3.2 3.3 BaS04 2.4 2.3 ZrSiO, 2.9 2.7 TiOo+AI2Os 9.6 11.8
a-Bu- n-IIeptsne
tune
13.8
13.8 8.7 3.3 2.3
2.2
__
B.E.T. method
N7 H20 16.1 A.l 11.8 A.2
13.8 8.6 3.2 2.4 2.8 9.5
13.8 11.7 4.2 2.8 3.5 12.5
.
n-Bu- n-Hep1150 tane tane 11.3A.2 56.GA.2 64.0A.a
10.5 8.8 3.2 2.1
2.7
13.8 2.7
13.8 8.7 3.6 2.4
9.5
a Aress in eq. metersln.
Table 111, Harkins and Jura. J . Am. Chem. Soc. 66, 1366 (1944) with a column of calculationa for the area8 by B.E.T. method using water isotherm with an assurned cross-section of 11.3.4,s for the water molecule.
c. &egg's Proposed Method. G h g g (18) has suggested a method very similar in theory to that developed by Harkins and Jura. It has, however, been applied for the most part only to charcoal and similar adsorbents having a very fine pore structure, and not, us yet, to materials for which an independent method of surface area appraisal is possible. It is quite probable, however, that his method when applied to low temperature nitrogen adsorption isotherms will yield surface areas in general agreement with those deduced by the B.E.T. or the Harkins and Jura plots.
111. LIMITATIONS AND GENERAL CRITIQUE OF THE GASADSORPTION METHODS 1 . Molecular Size In their original paper, Emmett and Rrunauer (4a) suggested obtaining cross-sectional molecular areas for the various adsorbates b y means of the equation
78
P. 11. EMMETT
where M is the molecular weight of the adsorbate, L is Avogadro'a number and d is the density of the solidified or liquefied adsorbate. This equation was derived on the assumption that the molecules have cross-sectional areas equivalent to those possessed in the liquid or solid forms and that they are close packed on the surface of the adsorbents. Experience has shown that, whereas this procedure gives very satisfactory agreement between areas obtained by molecules having similar boiling points (such as, for example, nitrogen, argon, carbon monoxide, or oxygen) it yields areas for other adsorbate molecules that may be as much as 50% smaller than those obtained by using nitrogen as adsorbate. Thus, the general conclusion drawn from work of Beebe, Beckwith, and Honig (20) from measurements with krypton is that a correction factor of about 1.3 has to be used to bring the area as measured by krypton up to the area as measured by nitrogen, provided the cross-section of the krypton atoms is obtained by applying eq. (13) to liquid krypton. Davis, Ile Witt, and Emmett (19) have similarly shown that correction factors of this same magnitude (1.2-1.5) have t o be used to bring into agreement the areas as determined by butane, Freon 21 (CHCLF), or butene with those obtained by nitrogen. The reason for this discrepancy is not as yet known. It seems that for some unexplained reason the molecules of these various adsorbates appear to have larger molecular areas than those calculated by eq. (13). 6. The k Value of Harkins and Jura (IS,14, 15) The method of Harkins and Jura avoids this difficulty by determining from the heat of wetting experiments on TiOz a surface area which in turn can be used to evaluate the constant k for various adsorbates a t given temperatures. Hy proceeding in this way they assign values of 3.83, 13.6, and 16.9, for example, to molecul6s of water a t 25"; n-butane, a t 0", and n-heptane, a t 25' C. These assignments enable them to obtain very good agreement among the area values for different adsorbates on a variety of solids. In Table 11, for example, in columns 2 , 3 , 4 and 5 are compared the areas that they obtained for six different solids by using nitrogen, water vapor, n-butane and n-heptane tts adsorbates. In Table I1 of one of their papers Harkins and Jura (13) have shown that the areas obtained by nssigning the cross-section areas to water, butane, and heptane on the basis of eq. (13) as recommended by Emmett and Brunauer (42) are too small by as much as 50%. Admittedly the molecular areas used by Harkins and Jura (13) produce a much better agreement for the surface area values on the different adsorbents than can be obtained by the B.E.T. method using the molecular areas obtained by eq. (13). It must be remembered, however, that the procedure of Harkins and Jura is tantamount to making the areas by nitrogen, butane, heptane, and water vapor agree on a given solidand
79
BURFACE AREA MEASUREMENTS
then applying these same relative areas to other solids. Columns 6, 7, 8 and 9 in Table I1 shorn that if a similar arbitrary procedure is followed in applying the B.E.T. method, agreement equally as good a s that of Harkins and Jura can be obtained for the various adsorbates. Accordingly, if one is using an adsorbate other than nitrogen one must be prepared to find that molecular areas calculated by eq. (13) may be as much as 50% too small to yield good agreement with the results obtained from the nitrogen isotherms.
3. ModiJicationsand Other Derivations of the B.E.T. Equation Cassie (21), and more recently Hill (22) have found that eq. (8), the B.E.T. equation, can also be derived by a statistical thermodynamic approach. Hill (23) has in addition pointed out that the original postulates of Brunauer, Emmett, and Teller (7) t,liat lead to the derivation of eq. (9) by a kinet,ic method can be shown by st,atistical thermodynamics to lead to the same equation. This is of interest since both Pickett (24) and Anderson (25) have suggested that the method of summation used in obtaining eq. (9) should be so altered as to produce an equation in the form V=
V,Cz(l -zn) (I+(C-1)a)
(1-2)
in which the symbols have the same meaning as in eq. (9). Equation (14) has the advantage of leading to the conclusion that when x= 1, the total amount of adsorbate in the adsorbent would be nV, rather than ( n + l > V , 2 0 1/Cn) that would be predicted by eq. (9). Fortunately, both eq. (9) and eq. (14) agree in the range of relative pressures used in calculating surface areas. It is only a t the higher values of x that they diverge. For this reason the critical comparison of eq. (9) and (14) need not in the present chapter be carried further.
+
4 . Con$rmution of B.E.T. Method on Nonporous Solids For nonporous solids the areas obtained by using nitrogen adsorption isotherms a t liquid nitrogen temperature in conjunction with “point B” method, with the B.E.T. equation or with the equations of Harkins and Jura all appear to check each other and to be in agreement with areas obtained by independent means. Space will not permit a complete listing of all comparisons that have been made by way of checking the validity of these methods involving the physical adsorption of gases. It will perhaps suffice to point out that good agreement is obtained between the gas adsorption method and areas estimated, (1) b y the heat of immersion method by Harkins and Jura (13) for TiOz, (2) by the electron microscope photographs for carbon black (Anderson and Emmett, 26; Emmett and De Witt,
80
P. H. EMMETT
27), (3) by the direct microscopic observations on sized glass spheres 7 microns in diameter (Emmett, 28), (4) by permeability measurements on ZnO pigments using various liquids as fluids (28), ( 5 ) by ultramicroscopic obscrvations on suspensions of these same ZnO pigments (28), and (6) by the adsorption of stearic acid and other molecules from solution on ZnO pigments (Ewing, 29). There seems to be little question that by the gas adsorption methods, reliable surface areas can be obtrtined for nonporous solid substances. 5 . Conjirmation on Large-Pore Solids
For porous solids in which the cracks, crevices and capillaries all appear to be many fold greater than molecular dimensions, there seems to be no reason for doubting the general applicability of the gas adsorption methods. It must, of course, be admitted that for such solids there are fewer methods available for obtaining independent estimates of the areas. By means of low angle scattering, however, it is possible to estimate the average particle size of gels and other porous catalysts. The few published results of such comparisons indicate that for porous solids, too, the nitrogen adsorption method gives reliable surface area values. Thus Elkin, Shull, and Roess (30) found that for a series of silica gels thc particle sizes revealed by low angle scattering measurements yielded area values which were larger than the areas obtained by the nitrogen adsorption method by factors ranging from 1.14 to 1.5 for the various gels. This constitutes excellent agreement in view of the combined uncertainties of the t\ro methods. A few data obtained by adsorption from solution may also be used t o test the arcas obtained by the adsorption method. Thus, Smith and Fuzek (31) found that the areas obtained by adsorbing palmitic acid from benzene on Raney nickel catalysts a g e d very well with thosc obtained by the nitrogen adsorption method. 6. Area Measurements on Charcoal and Other Fine-Pore Solids Only for materials giving isotherms approaching the flat Langmuir type shown in Fig. 4 does there appear to be much uncertainty relative to the surface area of the porous solid as estimated by the gas adsorption
method. Obviously when pores are sufficiently close to molecular dimensions, the very definition of surface becomes rather hazy. Fur example, the surface of some of the smaller capillaries will probably be reached only by the smallest adsorbate molecules. Hence, the area would depend upon the size of molecule used as a measuring unit. However, even for a silica gel sample that has pores so small as to cause its nitrogen adsorption isotherm to resemble those obtained on charcoal, the x-ray low angle scattering measurements of Elkin, Shull, and Roess (30) give good agreement with the nitrogen areas, But this is about the only means available for checking
81
SURFACE AREA MEASUREMENTS
the gas adsorption method when fine pores are involved, though a few meager comparisons have been made between areas deduced by the adsorption of nitrogen a t - 195"C. for charcoal, and areas obtained by adsorption from solution onto the charcoal. Measurements by Lemieux and Morrison (32) on the adsorption of homologous series of acids from aqueous solution yield surface areas that are 30 to 70% lower on gas adsorbent charcoal than those estimated from the nitrogen isotherms. As pointed out above, it seems fairly certain that V , values evaluated with the help of eq. (9) by the method of Joyner, Weinberger, and Montgomery (10) are as reliable as we know how to obtain at the present time for the surface accessible to molecules of this size.
7. Limitations and Precautions in Using the B.E.T. Method Other limitations and characteristics of the gas adsorption surface measurements have been pointed out in detail in previous publications (Emmett, 33,34). It will sufficehere to summarize them briefly. (1) The B.E.T. surface area method using nitrogen as adsorbate is generally applicable to any porous or nonporous solid in which a sample of 5 or 10 cc. has an area of at least 1 sq. meter. (2) By using adsorbates with low p , values such as ethane or ethylene at liquid nitrogen temperatures, it is possible to extend the sensitivity of the method sufficientlyto enable one to measure surface areas as small as 100 sq. cm. with an accuracy of about 10% (Wooten and Brown, 35). (3) The presence of a chemisorbed layer of gas on the adsorbent does not interfere with the low temperature nitrogen adsorption measurements provided, of course, that the pores are not so small that a monolayer of chemisorbed gas can block off or close up some of the tiny pores and hence actually decrease the amount of area to which the nitrogen has access. (4) In preparing the samples for measurement it is necessary to degas them at a temperature merely high enough to remove any easily condensible vapors such as water vapor which might be blocking part of the surface by being condensed in capillaries. The detailed procedure involved in malting the adsorption measurements is being omitted from this chapter because it has already beenpublished several times (Emmett, 28,33) and is, in addition,very little different from standard gas adsorption techniques that have been described in the literature for many years. OF SURFACE AREAMETHODS IN IV. APPLICATION OF SOLID CATALYSTS
THE
STUDY
1. Activity es. Surface Area At the outset it should be reemphasized that the low temperature gas adsorption methods measure the total accessible surface of a catalyst
a2
P. H. EMMETT
regardless of whether that surface is active catalytically, Accordingly, it does not necessarily follow that the total activity of a series of chemically similar catalysts will be proportional to the surface arerts as measured, for example, by nitrogen adsorption. Deviations from proportionality could arise from two different sources. In the first place, the fraction of the surface that might be in the form of active points or active planes could well be different on one catalyst than on another that is chemically identical. In such a case, of course, thc catalyst with the larger number of active points per cubic centimeter of catalyst will have the greater activity regardless of the absolute surface area as measured by adsorption methods. However, even if the surfaces of two catalysts that are being compared are completely uniform in activity and identical chemically, it still does not follow that the activities of the two catalysts will be identical. It is always possible that much of the surface on one of the catalysts is located in pores that, are so small as not to be readily available to the reacting gases. It could, therefore, easily happen that a catalyst with the larger surface as judged by the nitrogen method might possess the lower activity for catalytic reactions. Both Thiele (36)and Wheeler (37)have discussed this question of the relation between pore size and the fraction of the inner surface of the catalyst that can be utilized for a particular reaction. In spite of the above-mentioned factors that might interfere with correlations between surface areas and catalytic activity, it has often been found that a good correlation does exist. For example it has been pointed out in a paper by Owen (38) that the catalysts used in dehydrogenating butene to form butadiene have activities t h a t are directly proportional to the total surface areas. For such rnatcrials the surface area measurements may serve as a good control for prcdicting the activity that will be possessed by a given catalyst. 2. Measuring Surface Promoter Distribution Attent.ion has already been called to the use that has been made of surface area measurements in studying synthetic ammonia catalysts. I t has been possible, for example, to shorn that certain promoters have a specific influenceon the activity per unit surface area (4). Thus iron catalysts containing both KaO and A1203 as promoters have surfaces that are only about one-third as large as those containing only AlzOaas promoter and yet are several fold more active under synthesis conditions. Another valuable bit of information relative to promoters on iron catalysts has bcen made possible by t,he adsorption method for measuring surface areas. It has been established (Emmett and Rrunauer, 39,40)that the few per cent promoter content added to an iron catalyst actually covers as much as 50-75% of the surface of the reduced catalyst, thereby exerting
83
SURFACE AREA MEASUREMENTS
a much more marked influence on the activity of the catalyst than would be expected if the ratio of promoter atoms to iron atoms were the same on the surface as throughout the bulk of the catalyst. This observation in regard to the amount of promoter on the surface arose from the discovery that at
L vcc
60 .
p'
b.
vi
0 0
P
g 40 w 0 4In W
f,
-I
0
>
VAN DER WAALS CO
PURE IRON CATALYST
/
1
0
I
200
VAN DER WAALS CO
I
I
400 P. mm
600 ~~
~
81
FIG.6. Comparison of the isotherm for total carbon monoxide adsorption at -183" and for the physical adsorption on about 45 g. of pure iron synthetic ammonia catalyst (No. 973) and on a similar quantity of a doubly promoted iron catalyst (No. 931) (39).
- 183" C. carbon monoxide is quickly chemically adsorbed by the surface of a pure iron catalyst in amounts sufficient to cover the entire surface (4). Accordingly, it seemed reasonable to conclude that whenever the volume of CO chemisorption on an iron catalyst promoted with A 1 2 0 3 and KaO was smaller than the volume of nitrogen required to form a monolayer over the entire catalyst, one had an indication that part of the surface was being covered up by promoter molecules that were concentrating preferentially in the surface layer. For the doubly promoted catalyst, such measurements
84
P. H. EMMETT
indicated that only about 30-50y0 of the surface of the. iron was actually in the form of free iron atoms. Isotherms lending to such a conclusion are shown in Fig. 6. Fortunately it was possible (Emmett and Brunauer, 39, 40) to substantiate this conclusion in another way. Carbon dioxide st -78'
Fe -KgO - A l e 0 5 K . 0 r 1.59%
2ot-
=-z-r---
V A N DER WAALS COe
IRON CATALYST
I
0
1
I
I
200
400
600
P.mm
6( 0
ho. 7. Comparison of the isotherms for the total and the van der Wsals d o r p t i o n of carbon dioxide a t -78" on 45 g. samples of a pure iron Catalyst (No. 973) m d a doubly promoted iron synthetic ammonia catalyst (No.931) (39).
is only physically adsorbed by a pure iron catalyst; however, if any alkali is present a8 promoter in the catalyst a considerable amount of chemisorbed carbon dioxide is taken up by the catalyst. Isotherms illustrating this effect are shown in Fig. 7. A comparison of the volume of carbon dioxide taken up by the promoter molecules on the catalyst and the volume of CO chemisorbed by the same catalyst showed that the sum of the CO chemisorption and the COa chemisorption was approximately equal to the volume
SURFACE AREA MEASUREMENTS
85
of nitrogen required to form a monolayer over the entire catalyst. It, therefore, seemed reasonable to conclude that the CO was being chemisorbed by the iron atoms and that the CO? was being chemisorbed by the promoter molecules containing alkali. Accordingly, the combined chemisorption measurements of CO and COz together with a measurement of the total surface by low temperature nitrogen adsorption afforded a means of telling what fraction of the doubly promoted iron catalyst was covered with promoter molecules and what fraction consisted of iron atoms.
3. Area Measurements and Chemismplion The extension of the above described method for estimating the surface concentrations of promoters involves finding gases for the chemisorption measurements that are taken up by the principal catalyst component but not by the promoter molecules, or vice versa. This may not always be possible. Nevertheless, the results on iron catalysts are suggestive and point to a possible means of combining physical and chemical adsorption measurements to furnish more information about the composition of the surface of a catalyst than it had been possible to obtain previously. Indeed, area measurements facilitate the interpretation of all chemisorption runs on either single component or promoted catalysts. Thus Frankenburg (40a) and Davis (40b) found it convenient and helpful to know the surface area of the tungsten powder on which they were studying the activated adsorption of Hz and Nz respectively.
4. Application to Fischer-Tropsch Catalysts Another application that has been made of surface area measurements has to do with the preparation of Fischer-Tropsch catalysts. Anderson (41) and his coworkers at the Bureau of Mines are in the process of comparing the surface areas of the reduced and unreduced cobalt and iron oxide catalysts with the activities of the final catalysts. The work is in too early a stage of development to permit any final conclusions to be drawn because many factors other than surface area may enter into the performance of Fischer-Tropsch catalysts. Nevertheless, the surface area measurements will permit the control of one of the variables that otherwise would have remained an unkno\m complicating factor. Similarly, in work that has been carried out recently on the use of C14 and compounds of CI4 in studying the mechanism of the FischerTropsch synthesis over iron catalysts, the author and his coworkers have been guided continually by a knowledge of the surface area of the catalysts at every stage of the synthesis. Without such area measurements much of the work would have been impossible.
86
P. Ha EMMETT
6. B.E.T. Equation and Pore Size Measurements a. Work of Wheeler (42), of Shull (&), and of Ritter (4.9). Another application of surface area measurements has to do with pore size determination of porous catalysts. By assuming that the upper part of S-shaped isotherms such as shown in Fig. 1are combinations of monolayer adsorption and capillary condensation, it has long been customary to estimate the pore size distribution of catalysts by using the Kelvin equation
where u is the surface tension of the liquid adsorbate at temperature T of the experiment, V is the molecular volume of the adsorbate, p / p O is the relative pressure, R the gas constant, and 0 the angle of wetting of the adsorbate on the solid. The advent of the theory of multilayer adsorption and the establishment of the existence of such adsorption a~ a definite phenomenon has made pore size calculations of this simple type untenable and unreliable because part of the increased gas pickup is due to multilayer adsorption rather than to capillary condensation. However, within the last few years Wheeler (42), Ritter (43),and Shull (44) have proposed various methods for differentiating between the portion of an S-shaped isotherm due to multilayer adsorption and the portion that is to be attributed to capillary condensation. They have succeeded, therefore, in applying the Kelvin equation in such a way as to obtain estimates of pore size distribution in spite of the existence of multilayer adsorption. The details of the procedures followed separately by these three authors have not all been published. They have, however, all been discussed at summer conferences of the A.A.A.S. on catalysis. It will suffice, therefore, merely to outline the general procedure used by these authors and to point out that their work is being submitted forpublication in theverynear future. Wheeler (42) set up an equation in which the total amount of adsorbate held by the catalysts at any given relative pressure was visualized as being made up of that portion held by multilayer adsorption and another portion held by capillary condensation. Furthermore, he modified the KelviD equation slightly by postulating that the radius calculated by eq. (15) is not the radius of the empty pore but that of the hole left in the pore after the statistical number of layers predicted by the B.E.T. equation had been permitted to build up. Then by trying various Gaussian or Maxwellian distribution curves for the pores, he was able to deduce a pore distribution that would fit the adsorption data when both multilayer adsorption and capillary condensation ,were taken into consideration. Shull (44) carried out a similar treatment except that instead of using the B.E.T. equation to represent adsorption at high relative pressures, he estimated the multilayer
87
SURFACE AREA MEASUREMENTS
adsorption from a composite curve for nitrogen isotherms on a number of nonporous solids. He also improved the mathematical technique for deducing the pore distributions that will fit experimental isotherms of unusual shape. The approach of Wheeler and Shull to the problem is indirect in the sense that it merely confirms the agreement between theadsorption isotherms and certain pore distributions without showing necessarily that the distributions arrived at are the only ones that will give the observed
P/ Po
FIQ.8. Adsorption isotherms for argon, nitrogen, and butane on porous glaw No. 3. Open symbols, adsorption; closed symbols, desorption (45b). )
form of adsorption and desorption isotherms. Actually, these pore calculations are usually applied to desorption rather than to adsorption isotherms since it is generaIly believed that capillary condensation is more accurately measured by the Kelvin equation on desorption than on adsorption. b. Pore Size from Pore Volume and Surface Area, A less elaborate but frequently used method for estimating the average size of the pores of a
88
P. H. EMME'lT
solid makes direct use of surface area measurements. It is well known (Emmett and De Witt, 45a) that an estimate of the average pore size of a series of cylindrical capillaries can be obtained from the equation Radius
=
2v
2-
(16)
where V is the pore volume in cubic centimeters and A is the area of the porous adsorbent in square centimeters. As an illustration of this method reference may be made to the curves in Fig. 8 for the adsorption of nitrogen, argon, and butane on a sample of porous glass (Emmett and Cines, 45b). If the pore volume is calculated from the saturation adsorption value of each isotherm and the surface area is measured by the nitrogen adsorption isotherms in conjunction with eq. (8), one obtains average pore radii of 16.7, 15.6, and 15.5A. from the nitrogen, argon, and butane curves respectively of Fig. 8. These radii are in very good agreement with the one obtained (17.7A.) by applying eq. (15) directly to the nitrogen desorption isotherm and assuming that the radius calculated by eq. (15)is the size of the hole left after the capillary walls are covered with a single layer of adsorbed molecules. 6. Sintering Studies Finally, the surface area measurement method has proved to be very useful in following the changes in catalysts during use, during sintering, and during impregnation with promoters or additional catalytic components. Rics (46) and his coworkers have been especially active in exploring this application. As a matter of fact, the multilayer adsorption method was originally developed in the hope of being able to trace the relation between the change on sintering in the activity of aniron synthetic ammoniacatalyst as compared to the change in surface aren.
7. Present Slatus of the B.E.T. Method These few applications of surface area measurements of catalysts certainly do not exhaust the list of possible applications but they help to illustrate some of the principal ways in which such areas can prove nseful in a study of solid catalysts. They will doubtlessly be supplemented by many more examples as a larger number of workers begins to take advantage of the adsorption methods for following the surface areas and pore distributions of their catalytic materials. The author prefers to consider the B.E.T. and other methods for measuring surface areas by physical adsorption isotherms 8s still in the process of change and development. New suggestions, modifications and improvements in the method are to be expected. Nevertheless the measurement of catalyst areas by gas adsorption methods may be considered even
SURFACE AREA MEASUREMENTS
89
now to rest on a very solid foundation and may be used with confidence as a
tool in helping toelucidate the factors that influence theactivityof catalysts. REFERENCES 1. 2. 3. 4.
5. G.
7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.
29. 30. 31. 32. 33. 34. 35. 36. 37. 3s. 39. 40.
Tiislor, H. S.,A m . Scientist 34, 553 (1946). Bcnton, A. F.,J . A m . Chem. Soc. 48, 1850 (1929). de Boer, J. H., and Dippel, C. F., 2.physik. Chem. B3, 407 (1929). a. Emmett, P.If., and Brunauer, S., J . A m . Chorn. SOC.69, 1553 (1937). 1,. Brunauer, S.,and Emmett, P. H., J . Ant. Citent. SOC.69,2682 (1937). Russell, W.W.,and Taylor, H. S.,J . Ph!/s. Chrm. 29, 1325 (1925). Rent.on, A. F., and White, T. A., J . A m Cheni. Sor. 64, 1820 (1032). Bruiiauer, S., Emmett, P. H., and Teller, E., .I. Aut. Chent. SOC.60,309 (1938). Langmuir, I., J . A m . Chem. SOC.38, 2221 (1916). Langmuir, I., Phys. Rcv. 6,79 (1915). Joyner, L. G., Weinberger, E. B., and Montgomery, C. W., J. A k . Chem. SOC.67, 2182 (1945). Askey, P. G.,and Feachem, C. G. P., J . Sac. Chem. Znd. 67,272 (1938). Brunauer, S.,and Emmett, P. H., J . A m . Chem. SOC.67, 1754 (1935). Hiirkins, W.D.,and Jurn, G., J . A m . Chem. SOC.66, 1366 (1944). Harkins, U’.D.,and Jura, G., J . A m . Chem. Snc. 66, 019 (1911). Harkins, W.D., and Jura, G., J . Am. Chem. SOC.66, 1362 (1944). Livingston, H.K.,J . Chem. Phys. 12, 462 (1944). Emmett, P. I%, J . A m . Chem. SOC.68, 1784 (1946). Grcgg, S. J., Proceedings of the Conference on the Ultra-fine Structure of Coals and Cokes. British Coal Utilization Research Association, London, 1944,p. 95. Davis, R. T.,De W t t , T., and Emmett, P. H., J . Phys. and Colloid Chem. 61, 1232 (1947). Beebe, R..4., Beckwith, J. B., and Honig, J. bl., J . A m . Chem. SOC.67, 1554 (1945). Cassie, A. B. D., Trans. Faruday SOC.41, 450 (1945). Hill, T. L., J . Chern. Phys. 14,263 (1946). Hill, T.L., J . Chem. Phys. 14, 441 (1946). Pickett, G.,J . Am. Chern. SOC.67, 1958 (1945). Anderson, R. B., J . A m . Chem. SOC.68,686 (1946). Anderson, R.B., and Emmett, P. H., J . Applied Phys. 19, 367 (1948). Emmett, P. H.,and De Witt, T., Znd. Eng. Chem. Anal. Ed. 13, 28 (1941). Emmett, P. H., A. S. T. 111. Symposium on New Methods for Particle Size Determination, p. 95, 1941. Ewing, W.W.,J . A m . Chem. SOC.61, 1317 (1939). Elkin, P. B., Shull, C. G. and Roess, L. C., Znd. Eng. Chem. 37,327 (1945). Smith, H. A., and Fuzek, T. F., J . A m . Chem. SOC.68, 209 (1946). Lemieux, R. U.,and Morrison, I. A., Can. J . Research B26, No.5,440 (1947). Emmett, P. H., Advances i n Colloid Sci. 1, 1-36 (1942). Emmett, P. H.,Znd. Eng. Chem. 37,639 (1945). Wooten, L. A., and Brown, C. G., J . A m . Chem SOC.66, 113 (1!)43). Thick, R. W.,Znd. Eng. Chem. 31, 916 (1039). Wheeler, A, discussed at A.A.A.S. Conferences a t Gibson Island, June 1945. Owen, J. R., J . Am. Chent. SOC.69, 2559 (1947). Emmett, P. H.,and Brunauer, S., J . A m . Chem. Soc., 69,310 (1937). Brunauer, S., and Emmett, 1’. H., J. A m . Chem. SOC.62, 1732 (1940).
90 40.
41. 42.
43. 44.
45. 46.
P. H. EMMETT a. Frankenburg, W., J. Am. C k m . SOC.66, 1827, 1838 (1944). b. Davis, R. T., J. Am. Chem. SOC.68, 1395 (1946). Storch, H. H., Anderson, R.B., Hofer, L. J. E., Hawk, C. O., and Golumbic, N., U. S. Bur. Mines Tech. Paper No. 709, in press. Wheeler, A., discussed at A.A.A.S. Conferences at Gibson Island, 1945 and 1946, in press. Ritkr, H. L., discussed briefly a t A.A.A.S. Conferences a t Gibson Island, 1946. Shull, C. G., discussed by Dr. P. B. Elkin at A.A.A.S. Conferences a t Gibson Island, 1946; Shull, C. G., J . Am. Chem. SOC.70, 1405 (1948); Shull, C. G., Elkin, P. B., and Roess, L. C., ibid. 70, 1410 (1948). a. Emmett, P. H., and De Witt, T., J. Am. C k m . SOC.66,1253 (1943). b. Emmett, P. TI.,and Cines, M., J . Phys. and Colloid Chem. 61, 1248 (1947). Ries, 11. E., Van Nordst.rand, R. A., Johnson, M. F. L., and Bauermeister, H. O., J . Am. Chem. SOC.67, 1242 (1945).
The Geometrical Factor in Catalysis R. H. GRIFFITH The Gas Light and Coke Com.pany, L a d o n
CONTENTS I. Introduction . . . . . . . . . . . . . . 11. Reactions of Unsaturated Organic Molecules . . . . . . 1. Ethylene and Olefins . . . . . . . . . . . 2. The Hydrogenation of Other Unsaturated Organic hfoleculcs . . 111. Reactions of Acetylene . . . . . . . . . . . IV. The Synthesis of Hydrocarbons from Carbon Monoxide. and Related . . . . . . . . . . . . . Reactions . 1. The Fischer-Tropsch Process . . . . . . . . . . 2. Other Syntheses Rclated to the Fischer-Tropsch Process . . . . . , , V. The Formation of Aromatic from Aliphatic Hydrocarbons , VI. Thc Destructive Hydrogenation of Hydrocarbons . . . . . VII. The Rcactions of Hctcrocyclic Compounds . . . . . . . VIII. Thc Oxidation of Carbon Monoritlc . . . . . . . . IX. Future Developmmt . . . . . . . . . . . . References . . . . . . . . . . . . . . .
Page 91 92 92 95 95
97 97 loo 101 106 107 110 111 113
I. INTRODUCTION Many years before reliable information was available concerning the actual structure of any catalyst, attempts were made to consider the disposition of a reacting molecule on the active surface. So long as these conjectures were confined to reactants which were adsorbed by single atoms of the catalyst surface, little progress could be made by such hypotheses. Burk (1), however, studied the implications of two-point adsorption of diatomic molecules and came to the conclusion that some of the known facts concerning catalyst poisons could be accounted for by the assumption that this mechanism was involved. Balandin (2) shortly afterwards began the publication of a series of papers developed from the theory that the catalytic decomposition of a relatively large molecule could only take place by simultaneous adsorption a t several points; this was known as the Multiplet Theory. With cyclohexane, for example, attachment a t six centers mas supposed to be necessary before benzene could be formed by the loss of three hydrogen molecules. Calculations along these lines for metallic catalyst,s, whose lattice dimen91
R. II. GRJFFITH
92
sions were known, indicated that the distances involved were of the same order of magnitude as those of the C-C distances in benzene. Although the original form of this theory has been seriously criticized for several reasons (3), it provided a stimulus to further interest in problems of this type. Subsequent study of the relation between the dimensions of organic molecules, and of the catalysts which are effective in their conrersions, has proved most interesting and has estsblished the fact that twopoint adsorption frequently occurs. I t is with reactions of this type that this article is concerned; in particular, the importance of the geometrical factor in defining the propertics of a catalytic surface is considered. The reactions to which :Ittention will bc given are confined to a relutively narrow range, although others are likely to be found in due course which are suitable for similnr treatment.
11. REACTJONS OF UNSATURATED ORGANICMOLECULES 1. Elhylcne and Olcfins
The classical researches of Sabatier and his colleagues revealed the wide applicability of metallic nickel as a cataIyst for the hydrogenation of the olefinic double bond. Many years passed before the mechanism of this type of conversion was disclosed; this resulted from the discovery of deuterium, which made possible a study of the exchange reactions of ethylene. Because of its important bearing on geometrical problems, a brief account of this work must be included. When deuterium and ethylene are brought together in contact with a nickel catalyst, the exchange reaction CiH,
+ D i e CiHsD + HD
takes place. Two explanations of the mechanism of this exchange can be suggested, represented by the following series of changes. a. The associative mechanism, us proposed by Horiuchi and Polanyi (4), CITaD
CHn-CHD
LY I
in which the ethylene molecule is adsorbed on two adjacent nickel atoms by opening of the double bond, then reacts with an adsorbed deuterium atom to give a product adsorbed at one point only, but, by loss of a hydrogen atom, reverts to an adsorbed ethylene molecule containing 1). b. The dissociative mechanism due to A. cmd L. Farltus ( 5 )
93
QEOMETRICAL FACTOR IN CATALYSIS
in which single point attachment to the catalyst is involved. After many conflicting statements concerning these alternatives, the correctness of the associative mechanism was established by the work of Corm and Twigg (6a and b) which showed that no exchange takes place between C2H4 and C2D4 under conditions in which exchange between C2H4 and D2 occms. This proved that no free hydrogen or deuterium atoms were formed from the hydrocarbon alone. This conclusion was confirmed by further experiments with olefins of higher molecular weight. These revealed two very important facts: (1) that all the hydrogen atoms in the higher olefins, not only those adjoining the double bond, are capable of exchange with deuterium, and (2) that double bond migration can accompany the exchange reaction but only in the presence of hydrogen. These results would be expected from two-point adsorption of the olefin. Thus, with propene the following changes could occur: CH
CII-CHZ 7 I Ni N i
C€IS--CH-CHzD
+D __$
hlsorbctf molcculc
I
Ni Half-hydrogenated state 1
A;
--+I
CIiz-CH-CHzD
+ 11
I
Ni Ni hIolecule adsorbed in new position
,
CHj-CH-CTID Ni I
so that the desorbed molecule may be CHa.CH-=CHD or CHzD-CH=CHz. With butene, again, migration of the double bond could take place by successive addition and loss of a hydrogen atom, thus: CH,-CH-CII,-CHj
CII,--CH--CH:-CH,
I
Ni
I
Xi Butene-l
adsorbcd
+fI --f
I
Ni
C€I,--CH-CH-CHs -H --+
I
I
Xi N i Butene-2 adsorbed
Following these experimental investigations, Twigg and Rideal (7) considered the actual disposition of ethylene on the nickel lattice. This consists of a face-centered cube having a side of 3.50A., in which the 'closest Ni-Ni distance is 2.47A. If the C-C distance in ethylene is taken as 1.54A. and the Ni-C distance is assumed to be 1.82..3., the same as in nickel carbonyl, then the adsorbed ethylene molecule will fit on the close Ni-Ni spacing as shown in plan and elevation in Fig. 1, with a slight distortion of the Ni-C-C angle from the normal value of 109" 28' for the tetrahedral bond. It will not fit on the wider nickel spacing of 3.50A. (Slightly differ-
94
R. H. GRIFFITH
ent results will be obtained in this and similar calculations if the interatomic distances between metal and carbon are derived from the sum of the covalent radii for the respective atoms, this method giving alonger carbon-metal bond.) It is, therefore, evident that two-point adsorption of ethylene might occur, with slight distortion of the bond angles or some small displacement of the metal atoms from their normal lattice, on metal catalysts where the spacing is between about 2.4A. and 2.8A. In this group nickel, cobalt, copper, platinum, and palladium can be found and all of them can be used for the hydrogenation of olefinic double bonds. When a high concentration of olefin molecules is present, steric effects must also be considered. For example, the closest possible packing of ethylene molccules on the 110plane of nickel, results in no interference between adjacent molecules, and the whole surface should be nearly covered by the hydrocarbon. This is in accordance with the fact that ethylene can completely retard the conversion of para- to ortho-hydrogen on a platinum catalyst under conditions of relatively low temperature or high ethylene pressure (Farkas, Farkas, and Rideal, 8). With methylethylenes, interference between I % adjacent molecules becomes evident because of the large space occupied by the -CHa Fro. 1. The ethylene molegroups. The reality of this steric effect has cule adsorbed on nickel. The large circles represent carbon been demonstrated by Twigg and Rideal (7) atoms, the small circles nickel using the reaction Ha +Dae2HD a8 a means atoms. Bond lengths and anglcs of evaluating the amount of uncovered catare to but not alyst surface. They observed that a given diameters. catalyst could be completely covered by ethylene under conditions where much of it still remained bare in the presence of trimethylethylene. From the discussion above, it may be concluded that for reactions requiring two-point adsorption there should be a direct relation between the number of pairs of atoms in the catalyst surface at the correct spacing, and the catalytic activity. Confirmation of this effect has been provided by the work of Beeck, Smith, and Wheeler (9) on the hydrogenation of ethylene. As catalysts, they used films of nickel deposited by evaporation
-
GEOMETRICAL FACTOR I N CATALYSIS
95
onto glass under conditions where layers some 200-atoms thick were still fully accessible to the reactants. When the catalyst was prepared in a vacuum, the film was not oriented, but when nitrogen was present, the layer was completely oriented to expose the 110-plane in which the desired spacing occurs, and its catalytic activity per unit of surface was five times as great. Examination of this type of catalyst by electrop diffraction methods by the above investigators has contributed valuable additional evidence in support of the postulate of the geometrical relationships between reactant and catalyst. 2. The Hydrogenation of Other Unsaturated Organic Molecules Much of what has been said above can be applied to other hydrogenation reactions where the saturation of a double bond is involved. With many of these substances the steric factor must also be important, as the two-point attachment must often take place at or near the center of a long chain. With some reactions of this type it is probable that a departure from the normal nickel lattice may give increased activity, for it has already been noted that the 2.47A. spacing on nickel is not quite ideal; this applies still more if the C-Ni distance amounts to 2.0 A. rather than to the lonrer value of 1.82 A. derived from nickel carbonyl. Evidence has recently been provided by the work of de Lange and Visser (10) on nickel catalysts deposited on kieselguhr. This has established that the normal lattice of metallic nickel does not occur in materials prepared in this way, and that the reduced active catalyst is obtained as a result of the attack of hydrogen on a nickel hydrosilicate. By interaction of the diatomite and the nickel hydroxide deposited on it, an entirely new lattice is produced which provides a greatly increased total surface of catalyst, and which leads in the finished state to a widely dispersed and very stable arrangement of nickel atoms. It will be of great interest to obtain further evidence aa to the exact distance between pairs of nickel atoms in this catalyst, which is of exceptionally high activity.
111. REACTIONS OF ACETYLENE The reaction between acetylene and hydrogen, in contact with metallic catalysts, not only results in the formation of ethylene and ethane but also of products with much higher molecular weights. This latter process occurs at temperatures well below those required for the polymerization of acetylene alone, in the presence of the same metals. Sheridan (11) has studied the kinetics of the reactions involved and has thereby contributed to the understanding of the geometrical conditions which exist on such catalysts. Although the total disappearance of acetylene was largely
96
R.
a. GRIFFITH
independent of temperature in the rangc 0-130°, the proportions of ethanc, ethylene, or heavier hydrocarbons varied. the simpler hydrogenation reactions having lower apparent activation energies. These differences between the behavior of 'ethylene and acetylene, on hydrogenation with simple metallic catalysts, probably arises from the greater unsaturation of acetylene. The adsorbed ethylene molecule, held a t two points, probably has the carbon-carbon skeleton of ethane; acetylene similarly held has that of ethylene and the C-=C distance is therefore 1.33A. instead of 1.54A. for C-C. The hslf-hydrogenated Rtate for acetylene therefore has the structure CH==C:HZor --CII--CH,. In the I,zt,ter, one of thc carbon
I
1
1
Xi Ni Xi atoms has an unpaired elcctron which c ~ ~ i lead l d to its taking part in a polymerization reaction. Sheridan hiis considered all three possible processes from the geometrical viewpoint. Assuming the C-Ni distance t o be 2.0 A, the C-Pd distance to be 2.1 A. , the C-Pt distance to be 2.1A., and the angle M-C=C to be 121°, the required spacings for two-point adsorption of acetylene become 3.33 A. for nickel, and 3.43 A. on palladium or platinum; the spacings for ethylene adsorption would similarly be 2.87 and 2.94 A. Those actually available are 2.48 and 3.5 A. for nickel, 2.75 and 3.88A. for palladium and 2.75 and 3.91 A. for platinum. Some strain, therefore, is required to fit ethylene to any of these lattices, and in an opposite direction to fit acetylene on the long spacing. On purely geometrical grounds, the adsorption of ethylene on the more abundant short spacing might be expected to outweigh that of acetylene; in fact, however, acetylene is more strongly adsorbed, and it is evident that additional factors must be involved. The hydrogenation of acetylene may therefore be explained by a mechanism very similar to that already discussed for ethylene, and of course the two reactions can go forward simultaneously on the different spacings of the same catalyst. The formation of higher hydrocarbons can also be explained by similar postulates, as follows. The characteristic feature of the adsorbed acetylene molecule is that it is still unsaturated, whereas adsorbed ethylene is not. Moreover, it is important to note that the polymerization reaction with acetylene takes place when hydrogen is present but does not occur in its absence; it therefore seems likely that the half-hydrogenated state of acetylene plays a vital part in the process. For ethylene, the corresponding intermediate is CH~-CITB, but for
I
acetylene it must be either CH=-CI12 or -CH--CH2.
I
Ni
I
Ni
I
Ni
Ni In the latter, t,here
GEOMETRICAL FACTOR IN CATALYSIS
97
is a carbon atom with an unpaired electron which may take part in a chain reaction with other adsorbed acetylene molecules and produce hydrocarbons of various molecular weights, due to the presence of molecules in the halfhydrogenated state. The disposition of the polymer on the nickel catalyst is different from that required for the initial acetylene adsorption, and the process may be expected to continue when both the long and short Ni-Ni spacings are available, as they are on the 110-plane of nickel. Whether the building up of the polymer requires simultaneous adsorption of several carbon atoms or not is uncertain, but this particular matter is not of primary importance. It does, however, appear to be most probable that the half-hydrogenated acetylene fragment held on the 2.47 A. pair of nickel atoms will react with the terminal pair of a larger olefin molecule held on an adjacent pair of the same spacing, and there is no need to suppose that the whole of the larger chain is held. This view is supported by the fact that the branched paraffins which are formed contain only single methyl groups in the side chains. It is not justifiable to carry these conjectures any further without much more experimental evidence, but the results thus far available indicate that adsorbed ethylene has the ethane skeleton while adsorbed acetylene has the ethylene skeleton, and that the differences in their behavior can be explained on this basis.
IV. THESYNTHESIS OF HYDROCARBONS FROM CARBON MONOXIDE, AND RELATED REACTIONS
I. The Fischer-Tropsch PTOC~SS Metallic catalysts of the iron-cobalt-nickel group again come under consideration when evidence is brought forward as to the geometrical requirements of the Fischer-Tropsch synthesis. Although many features of this process have no direct bearing on the geometrical study of the catalysts which are used, it is necessary to give a brief summary of their salient points in order to present all the evidence in proper perspective. The work of Craxford (12) and of Herington (13)has shown that when carbon monoxide and hydrogen are passed over catalysts consisting essentially of metallic cobalt, a series of reactions takes place which may be represented by the following scheme: a. Carbon monoxide is adsorbed by the metal, and in this state reacts with hydrogen, producing water and a surface layer of cobalt carbide.
co
C
b. More hydrogen then reacts with the carbide to form adsorbed groups held by the surface cobalt atoms. If the conditions are
-CHn-
98
R. H. QRIFFITH
such that there are numerous adsorbed hydrogen atoms also present, then these -CH2groups will be converted into methane 'and no synthesis of higher aliphatic hydrocarbons will take place. More favorable results, however, arise when relatively few hydrogen atoms are adsorbed, and when the catalyst surface is almost completely covered by hydrocarbon fragments or carbide. c. Under these conditions, two adjacent -CHs fragments combine to
I
CO
produce adsorbed ethylene held on two adjacent cobalt atoms. d. Growth of the hydrocarbon chain then occurs by reaction of a further adsorbed -CHzfragment with the ethylene molecule, in a manner similar to that involved when deuterium exchanges with ethylene, and to that suggested for the production of higher hydrocarbons from acetylene. Intermediates of the type C€Iz-CH = CH, can be readsorbed
I
co on two Co atoms, by migration of hydrogen, in the form CH2-CH-CH3.
I
I
c o Co e. This process can be repeated, the double bond generally migrating to the terminal pair of carbon atoms so that the hydrocarbon formed usually has a straight chain. The chances of reaction of a -CHr fragment with a larger molecule held by a pair of carbon atoms other than those a t the end of the chain arelesslikely, but this process will sometimes take place and will give rise to an isomer in which the branch consists of a single methyl group. Equal growth of two branches to any considerable length is unlikely to occur on steric grounds. I. Desorption of the hydrocarbon will thus produce a 1,2 olefin, or, following hydrogenation of the double bond, a p a r f f i . g. Cracking of large hydrocarbons, sometimes accompanied by hydrogenation, may also take place on the same catalysts. Apart from the conclusions derived from a study of the direct synthesis, certain other factors support the correctness of this picture of the geometrical relations between catalyst and reactants. If the interatomic distance and bond angles were not important in controlling the synthesis, it might be supposed that the adsorbed -CHZfragments all link together into a few giant molecules covering nearly the whole surface, and that hydrocarbons of varying molecular weight arc then produced by splitting of the macromolecules. But such a picture would imply that most of the C-C and the C-Co bond angles were go", would leave no room for the attachment of hydrogen
99
GEOMETRICAL FACTOR I N CATALYSIS
atoms, and would be expected to produce many types of branched-chain hydrocarbons. To reject such a picture does not mean that large aliphatic hydrocarbon molecules cannot cover a great deal of the surface. As an example, Fig. 2 shows a t A the projection of a CISchain which is adsorbed by two
0
a
a
a
z
a
0
A
a
a
0
:>:
a
0
0
a
a
B
0
a
0
a
0
0
.
-
a
0
0
I t
FIG.2. The disposition of a CISchain on a metallic cobalt catalyst. A - attachment by two distant pairs of carbon atoms, with little departure from the planar zigzag chain. B - attachment to six cobalt atoms in onc row, with extensive departure from the planar chain. Cobalt a t o m shown as full black circles.
pairs of carbon atoms on two pairs of cobalt atoms in a 100-plane of the metal catalyst. Some departure from the planar form of hydrocarbon chain is necessary to give this arrangement, but the normal bond angles can be retained. Such an attachment effectively blocks much of the surface, and several similar molecules could pack closely side by side. I n actual practice, Fischer catalysts are well known to suffer from gradual
100
R. H. QRIFFITH
deterioration by the building up of waxy hydrocarbons, which are not normally desorbed but which can be removed from the surface by treatment with hydrogen. The mechanism of this cleaning process is apparent from the diagrams of Fig. 2, as the attack of hydrogen atoms can take place in a system like this and will result in desorption of the hydrocarbon. It is even possible to imagine a long-chain paraffin molecule adsorbed at many points so that every cobalt atom in a row is occupied, as in B of Fig. 2. This arrangement requires repeated Cs units to fit in place. The chances of such an arrangement are much smaller than those of adsorption as shown in ,Fig. 2 (A) where so many more centers are available. Similar arguments apply t o the 110-plane of a cobalt or nickel catalyst, in which the atoms are arranged in parallel rows, with a row to row distance of 3.5 A. and an internal row spacing of 2.5 A. One further piece of'evidence can be brought forward supporting the chain growth hypothesis. Since adsorbed ethylene is present, according to this view, it should be possible to incorporate this hydrocarbon into the larger molecules built up in the synthesis, by adding it t o the reacting mixture of carbon monoxide and hydrogen. That this can be done was demonstrated at an early date by Smith, Hawk, and Golden (14). On the other hand, acetylene added to the gas should not take part in the synthesis, as it can only be adsorbed on the wider epacing of cobalt or nickel atoms, and enough hydrogen to convert it to ethylene is extremely unlikely to be present in the adsorbed state on adjoining metal atoms. Herington (13) has shown that this predictipn is correct and that acetylene is not converted into higher boiling products during the synthesis. 2. Other Syntheses Related to the Fiseher-Tropsch Process Comparatively little is yet known of some synthetic reactions which obviously resemble the Fischer-Tropsch process very closely, but they are worth brief mention because they are als? likely to be controlled by geometrical factors. The 0x0 synthesis (15) of aldehydes by the interaction of ethylene or other olefins with carbon monoxide and hydrogen is carried out in contact with cobalt' catalysts at temperatures in the range 110-150°,and under a pressure of 100-200 atmospheres. Cyclic olefins react similarly; for example, cyclohexene giveH hexahydrobensaldehyde. There can be little doubt that a two-point adsorption of the hydrocarbon must take place and that the adsorbed molecule then reacts with carbon monoxide and hydrogen; the difference between this process and that responsible for the normal hydrocarbon synthesis is that adsorbed carbon monoxide survives as such under the less drastic temperature conditions which are employed. Owing to the fact that a variety of isomeric aldehydes are produced, this system deserves further detailed study on geometrical lines.
GEOMF2RICAL FACTOR I N CATALYSIS
101
A modified type of hydrocarbon synthesis, known as the "Iso-synthesisll' also requires much more attention because the catalysts here employed are oxides and not free metals (16). Mixtures of thoria, alumina, and zinc oxide are used at about 450" under high pressure, and produce good yields of 2-methylpropane; at higher temperatures naphthenes and aromatic hydrocarbons also appear, while a t lower temperatures isobu tyl alcohol can be isolated. By passing dimethyl ether and hydrogen over the catalysts, hydrocarbon products are obtained similar to those from CO and Ha. The ether is therefore likely to be an intermediate. More will be said about the structure of oxide catalysts in the following sections. V. THEFORMATION OF AROMATIC FROM ALIPHATIC HYDROCARBONS I n comparison with the information about geometrical factors €or metallic catalysts, little parallel progress has been made with oxides or sulfides. This is largely due to the uncertainty which exists concerning the precise chemical composition of these catalysts in their active state, and partly to the lack of reliable data for the crystal structures of some of the lower oxides and sulfides. However, there are one or two instances where enough is known about catalysts of these types to justify some attention at this point. Some of the most thoroughly investigated cases constitute the group of catalysts used in the ring-closing reactions which lead to the production of aromatic hydrocarbons from paraffins or olehs. These consist of oxides of vanadium, chromium, and molybdenum, or of complex and supported catalysts containing one of these oxides. Herington and Rideal (17) have studied a series of catalysts containing molybdenum, which were prepared by the reduction of MOO,, of supported MOO,, and of complex molybdates. These had activities ranging from zero for the plain oxide to high values for some of the complexes and for Mool supported by alumina. It was found that the plain oxide could be reduced at temperatures above 500", while the supported catalyst remained in the form of oxide; similar differences were disclosed among the complex molybdates. Ammonium cupromolybdate, for example, was reduced to metallic molybdenum and was inert, while other compounds, such as vanadomolybdates, although they decomposed, led to catalysts in which Mooz persisted. These facts suggested that the oxide Moo2 was essential in bringing about the ring-closing reaction, and the special features of its crystal lattice were considered. This is of the rutile type (body-centered tetragond system) and is shown in Fig. 3, together with a plan of the 110-plane in Fig. 4. Here the shortest distance between molybdenum atoms is 2.798., the next
102
R. H. GRIFFITH
closest spacing being 3.71A. For a two-point adsorption of an olefin on adjacent molybdenum atoms at 2.79A., the Mo-C-C angle works out at 108’ instead of the tetrahedral angle; only slight distortion is therefore necessary. Molybdenum metal, on the other hand, is a body-centered cubic crystal with an edge of 3.14A. and the shortest Mo-Mo distance 2.2A.; twopoint adsorption on this lattice was considered less likely than on the oxide. In view, however, of what is said later about the properties of molybdenum disulfide, it does not appear that this explanation of the inactivity of the
0 . 0
FIQ.3. The unit cell of Moon Molybdenum atoms, full black circles Oxygen atoms, open circles a-4.86A. b =4.86 A. c -2.79 A. d =3.71 A.
. .
0.0
0.0 0 . 0 IR c _
FIQ.4. Plan of the 110-plane of MoOa. Molybdenum atoms, full black circles Oxygen a t o m , open circles
metal is satisfactory, and some other reason must exist. It may, for example, be due to the absence of oxygen atoms, or to a drastic shrinkage of the surface of the catalyst on conversion of the oxide to the metal. Some other features of these ring-closing catalysts are also interesting from the geometrical aspect. If it is assumed that the formation of an aromatic hydrocarbon can only occur when the terminal pair of carbon atoms in a chain is adsorbed, then it is possible to calculate the probability for the conversion of various initial aliphatic hydrocarbons into aromatics; these probabilities are proportional to the rates of conversion under a given set of conditions. The results of such calculations (17) show excellent agreement with experimental figures, and this agreement between the two is strong evidence in support of the argument about the importance of the spacing on the catalyst surface.
QEOMETRICAL FACTOR IN CATALYSIS
103
Besides molybdenum oxide, chromium oxide catalysts are very important for the production of aromatics from open-chain hydrocarbons. Certain features of this series of catalysts still remain obscure, but it is already highly probable that their behavior may be explained on geometrical grounds. The important facts about these chromium oxide catalysts are as follows: CrTO, gel, prepared by precipitation of chromic salts by ammonia, yields a highly active catalyst on treatment with hydrogen. This activity is not retained on prolonged use, and falls off rapidly when the material is regenerated by oxidation; this decay is due t o conversion t o crystalline CrzOa (Visser, 18) which has been shown to be inactive (Turkevich, Fehrer, and Taylor, 19). But if the chromium oxide is deposited on alumina, it retains its original high activity and even survives regeneration by oxidation at quite high temperatures. Moreover, the pretreatment of the alumina makes an appreciable difference to the activity of the catalyst, although A1203 itself is quite inert as a catalyst for the ring-closing reaction. Figure 5 shows the
I
I
Mx)o
I
I
mo 900.
,
11000
T E M P ERATU R E
FIG.5. The activity of chromium-oxide-alumina catalysts for production of aromatic hydrocarbons. Effect of teniperature to which y-A1,O1,H20 has been heated before application of CrOs.
activity of chromium oxide-alumina catalysts in relation to the temperature to which the hydrated alumina has been heated before deposition of the chromium oxide (Grifith and Morcom, 20). I n Fig. 6 is shown the effect of temperature on the conversion of y-AlaOa to a-A1203, these results being due to Biltz (21). These together provide strong evidence that the alumina is most effective in the anhydrous y-form. It is, therefore, very likely t h a t the active chromium oxide catalyst exists in the form of y-CraOs, with a crystal lattice closely resembling that of y-Fe20s or y-Alz03,b u t not capable of existence by itself under the conditions of the catalytic conversion.
104
R. H. ORIFFITH
Visser (18) has attempted to identify this form by X-ray examination, but has not been successful; he has, however, shown that a pure Cr003catalyst heated at 475' for 1hour has a surface area of 28 m.a/g., while a preparation containing 30% Alu208 had a surface of 168 m.2/g. In view of these observations it is also very likely that ring-closing catalysts obtained by depositing
700.
900.
11008
1300'
T E MPERATURE Fro. 6. The conversion of y-AlnOa into a-AlnOa. The transition temperature coincides with the point in Fig. 6 where activity declinee after p d n g the maximum.
MooI and VaO3 on alumina, the latter being described by Kazansky et al. (22), also function in a crystalline form isomorphous with -y-Alz03. The lattice of Y-AlaOa or -y-FelOa is of the spinel type; Fig. 7 shows a plan of the 100-planefor a crystal of this series; a similar arrangement exists in several other planes. Some vacant metal atom spaces occur at unknown positions. It is evident that the metal atoms are here present in rows, and that the spacing is suitable for two-point adsorption of a hydrocarbon chain. Moreover, the wide spacing between parallel rows of metal atoms and the presence of two intermediate rows of oxygen atoms would be expected to diminish the tendency for other reactions to take place, and little choice is left to the adsorbed hydrocarbon chain except to form a ring. The conditions are not quite so severely limited in other planes of the crystal, and, in fact, all these catalysts suffer from side reactions which produce heavy polymers; the removal of these substances makes regeneration necessary. I n the above a r g k e n t , no attention has been given to the fact that hydrogen must be split off from the aliphatic hydrocarbon during its conversion. There is no evidence yet as to how this takes place, but it is possible that the oxygen atoms in the catalyst play some part in this process, and a mechanism like this would explain the inactivity of metallic molybdenum.
105
GEOMETRICAL FACTOR IN CATALYSIS
Ring closure with a sulfided catalyst has been described by Hummer and Taylor (23). When molybdenum dioxide supported on alumina mas treated with sulfur compounds it was found to retain some of the sulfur and to gain in activity as a catalyst: It was, however, quite distinct in its properties from a normal MoSa catalyst which has a strong cracking action. 0
0
0
0
0
0
0
0
0
0
0
0
0
0
..
0
0
0
0
0
0
0
0 0
0
0
0
0 0
0 0
111
FIG.7. Plan of the 100-plane of -pALOs. Aluminum atoms, full black circles Oxygen atoms, open circles Dimensions for y-Cr201would be very similar. The lines of active atoms are widely separated.
It is therefore probable that the sulfided catalyst retains a spinel structure, in which partial replacement of oxygen by sulfur has occurred. The fact that r-AlaOa itself is not a catalyst for ring closure is important, as it illustrates the great significance of the chemical character of the element concerned in determining the presence or absence of catalytic activity. Herington and Rideal (24) have considered what products may be formed from a large variety of aliphatic hydrocarbons, on the assumption
106
R. H. GRIFFITH
that two-point adsorption always takes place but can often do so in more than one way. They have been able to give a satisfactory explanation of all the results which are obtained in practice with a catalyst consisting of chromium oxide on alumina, including those reactions which involve isomerization and the formation and rupture of secondary rings. The success of such approaches to the problems involved emphasizes the importance of studying thoroughly the structure of the catalysts used.
VI. THEDESTRUCTIVE HYDROGENATION OF HYDROCARBONS A great many substances have been described as catalysts for the conversion of heavy hydrocarbons into those of lower boiling point, by treatment with hydrogen undcr pressure. Most of the publications on these processes are not concerned with the precise chemical composition of the catalysts but only with their rough classification into groups of the periodic system, so that it is not possible to gain much information from them which is useful for the present discussion. There is, however, no doubt as to the exceptional activity of certain molybdenum compounds for this type of reaction; the disulfide is outstanding and the dioxide is also highly effective. What has already been said about the mechanism of ring closure on MoOa will have indicated that this oxide should be capable of adsorbing almost any olefin, and it could behave similarly towards those with a cyclic structure in suitable temperature ranges. The wide applicability of this catalyst is, therefore, no more surprising than that of metallic nickel for the hydrogenation of the olefinic double bond. The molybdenum catalysts, unlike nickel, are not poisoned by sulfur compounds, and when such compounds are present in the hydrocarbon undergoing treatment, partial conversion of Mooato MoSz will take place. For this reason, as well as because of the particularly high activity of the disulfide, it is of interest to consider the structure of MOSSwith reference to the adsorption of organic molecules on its surface. Molybdenum disulfide consists of a layer lattice with hexagonal symmetry. Each plane of molybdenum atoms is held by covalent bonds between two planes of sulfur atoms; part of the structure is shown in Fig. 8. The important feature of this lattice is that the molybdenum atoms are arranged triangularly 3.15A. apart. Such planes could readily be exposed in a crystalline MoSa catalyst from which one layer of outer sulfur atoms had been removed by reaction with hydrogen. On this lattice, with the Mo-C bond 2.14A. it can be seen that two-point adsorption of an olefin would bonds. This fit is not take place with an angle of 112' for the Mo-C-C
GEOMETRICAL F.4CTOR I N CATALYSIS
107
quite so good as on molybdenum dioxide and therefore does not explain the greater activity of MoS2. It must be remembered that an important step in the destructive hydrogenation process is a reduction in the molecular weight of the hydrocarbons. This involves rupture of a C-C bond. Now, although the mechanism of this breakdown is probably very similar to that of hydrogenation of a double bond, involving two-point adsorption of adjacent carbon atoms, it is more exacting in its requirements of spacings on the catalyst surface. If the adsorbed molecule is a paraffin, its attachment to the surface involves simultaneous loss of two hydrogen atoms and four adjacent sites on the catalyst are required; only if two other hydrogen atoms are also available on other adjacent sites will the C-C link be broken and two separate molecules desorbed by taking up two hydrogen atoms each. The whole operation therefore requires six adjacent sites. When there is a deficiency of hydrogen atoms, the larger molecule may be desorbed only as unchanged paraffin, or may even lose hydrogen to produce an olefin. This reasoning suggests that the higher activity of MoS2 can be attributed partly to the greater number of molybdenum atoms in close proximity compared with those in the best plane of Mooz. Perhaps also differences in the '. The MoSz showing incomplete cells with molybdenum activation energy of hydrogen adsorp- layers exposed. t i o n a r e i n v o l v e d . T h e o b serv ed behavior of a catalyst whose character is largely dependent on geometrical factors with respect to one reactant may therefore be equally dependent on the adsorption of a second reactant which is held by single point attachment. VII. THEREACTIONS OF HETEROCYCLIC COMPOUNDS Calculations similar to those given above may be extended to organic compounds other than hydrocarbons; for example, in the series furan,
108
R. H. ORIFFITH
thiophene, and pyrrole, hydrogenation catalysts can be found which will lead either to the formation of the corresponding tetrahydro derivatives or to much more extensive breakdowns with removal of the heteroatom as water, hydrogen sulfide or ammonia. Reference to the bond angles and interatomic distances for these molecules shows that as far as the carbon atoms are concerned their values are little different from those of the olefin molecule. I t wouldbe expected, therefore, that catalysts which were effective for hydrogenation of the latter would also function with the heterocyclic molecules. This is found to be so, in that nickel catalysts are known to give tetrahydrofuran and pyrrolidine by the hydrogenation of furan and pyrrole at 180" (Padoa, 25). Tetrahydrofuran is also formed by the use of platinum (Starr and Hixon, 26), asmium, or pdttciium (Shuikin, Nikiforov, and Stolyarova, 27) as the catalyst, and pyrrolidine is similarly produced by palladium or rhodium catalysts (Zelinskii and Yurev, 28). In all these metals there are spacings of the atoms very similar to those in metallic nickel, the hexagonal osmium lattice having a equal to 2.71 A. The corresponding change with thiophene does not take place because the catalysts are poisoned by the sulfur. In order to hydrogenate thiophene, therefore, sulfur-resisting catalysts must be sought. These have been found in the sulfides of molybdenum and nickel, described by Moldavskii and Kumari (29) and by Cawley and Hall (30). In both these series of experiments, static systems under pressure were used, so that no sulfur was carried away from the reaction zone and at least part of the catalyst remained in the form in which it was originally added. The structures of hloSr and NiS are therefore to be considered. What has already been said about the disposition of olefins on MoSz can obviously be applied equally to the thiophene molecule, assuming that one double bond is opened by two-point adsorption. ThiR argument, which can explain the known activity of MoSa for this conversion, would also indicate a similar activity for NiS, since this sulfide consists of a hexagonal lattice with alternate layers of nickel and sulfur atoms in which the shortest Ni-Ni distance is 2.65A., and which wduld, therefore, bz suitable for two-poin t adsorption of thiophen e. When the reactants are flowing continuously over the catalyst, however, nickel sulfide no longer remains effective because NiS is not stable in contact with hydrogen at temperatures above 150'. It is changed into the subsulfide, NisSz,which has a rhombohedra1 structure and a unit cell edge of 4.04A. (Wrstgren, 31). No pairs of nickel atoms exist on the 100-plane of this crystal at or near the correct spacing to give two-point adsorption of thiophene. On the lll-plane, groups of tliree nickel atoms are present, as shown in Fig. 9, in a triangular arrangement, and any pair of these would
GEOMETRICAL FACTOR IN CATALYSIS
109
give the correct distance for the required adsorption of the organic molecule, but do not provide for the additional two hydrogen atoms also needed. Nickel subsulfide, therefore, is an effective catalyst for the hydrogenation of carbon disulfide and other simple organic sulfur compounds which are held
@
0
O
O
0
0
0'
00 00 00 0 00 00 00 0 00 00 00 0
-11 0 0
0
. 0
0
0 0
0
0
.
0
0 0
.
0
0 0
. 0
.
.
0
0
0
0
0 0 F m 9. NickeI subsulfide. Plans of the 100-plane (upper) and of the Ill-pkne (lower).
by single-point adsorption (Crawley and Griffith, 32) but not for the destructive hydrogenation of organic molecules. These observations emphasize the fact that direct measurement of the separate adsorption of thiophene and of hydrogen on this catalyst would not disclose this important feature. Determinations of this kind must be used with caution in the prediction of results to be expected when multipoint adsorption is involved.
110
R. H . QRIFFITH
I n spite of the fact that nickel subsulfide does not produce tetrahydrothiophene, it is possible to decompose thiophene into hydrogen sulfide and butene by passage over this catalyst at about 400" . The mechanism of this breakdown, however, is entirely different from any of those discussed above and depends on the reaction of the sulfur atom with metallic nickel which is formed when nickel subsulfide is treated with hydrogen at temperatures above 350' (Griffith, Marsh, and Newling, 33). But a geometrical factor may also be involved even in this breakdown of thiophene. While this reaction proceeds smoothly in pure hydrogen, the conversion is completely inhibited in coal gas. The reason for this failure is not that the simpler olefins in the coal gas are competing with the thiophene, but that cyclopentadiene is strongly adsorbed and polymerized, to the exclusion of the far less plentiful thiophene. Several other substances which are only held by single-point attachment are also contributing to the covering of the surface; for example, carbon disulfide has been found to retard the conversion of thiophene in hydrogen. In studying this series of changes, it is interesting to note that the heteroatom appears to have little influence on the alignment of the molecule. This is not always so, and the same group of heterocycliccompounds can take part in an entirely different set of reactions such as: CAHS +NH,-+CdHJH +H*S C,H,O+H,S -+C4HIS G O ,
+
similar changes occurring rather more readily with the tetrahydro derivatives in contact with alumina catalysts a t 400450' (Yurev, 34). It is possible that two-point attachment of the heteroatom and of one carbon atom may be involved in this type of conversion.
VIII. THEOXIDATION
OF CARBON
MONOXIDE
None of the systems so far considered has involved oxygen as one of the reactants, and in fact very little is yet known about the effect of geometrical factors in oxidation. One example can be found in the Hopcalite reagents and catalysts described by Pitzer and Frazer (35). They have shown that the higher oxides of cobalt, nickel and manganese, which are effective for oxidation of carbon monoxide, have all metal-oxygen bond lengths very similar to one another; thus for MnOz the bond length is 1.84A., for Ni203, 1.84A., for CozOs, 1.85A., and for Co30,, 1.75A. Copper oxide with a bond length of 1.87 A. is not itself active but functions as a promoter. Other important features of the process are (1) that carbon monoxide can undergo activated adsorption on other catalysts which do not effect the oxidation reaction, (2) that carbon monoxide can be oxidized on inactive
111
QEOMETRICAL FACTOR I N CATALYSIS
forms of manganese oxide but the process is not catalytic, and (3) that poisoned catalysts have increased coordination round the metal atom. These facts indicate that the carbon monoxide must be held simultaneously by both metal and oxygen atoms. Thus, the adsorbed molecule must be represented as C 0 -c 0 0-M
' -b
and not as O-M--O--CO
I
or 88 0-M-0.
After the loss of COathe manganese oxide is reoxidized. In any further study of the geometry of these surfaces, it would be instructive to determine the exact crystallographic species involved in the truly catalytic stage. Alekseevskii and Frid (36) state that only the a-form of MnOz is active for the oxidation of carbon monoxide, and recent work by Feitknecht and Marti (37)has provided new knowledge concerning the higher oxides of manganese and the manganites. For example, they have shown that manganous manganite and the manganites of other divalent metals have a double layer lattice in which planes of MnOa with hexagonal symmetry are separated by disordered layers of the hydroxide of the lower valency metal. The disorder of the intermediate'planes lies in their incompleteness, and a very small proportion of these hydroxides is sometimes sufficient to stabilize the whole structure. It seems probable that the unique properties of the Hopcalite catalysts are dependent on this structure. Interpretations rather similar to those described above may be placed on results observed by Eckell (38). Carbon monoxide was oxidized by contact with catalysts consisting of ferric oxide to which varying amounts of alumina had been added. It was shown that a quantitative relationship existed between the interatomic distances in the catalyst, which decreased as the content of A1203 increased, and the activation energy for the oxidation. Beyond the miscibility gap at 25% &03, both retained a steady value. It is possible, however, that in a system,of this type, where reduction of FeaOoto Fe304is known to occur to a recognizable extent, the effects depend on factors other than the actual spacing of the reacting molecules on the surface. The subject is one which lends itself to extended investigation and is closely related to the problems of promoter action. DEVELOPMENT IX. FUTURE It will be evident, from the limited number of items discussed above, that the study of the geometry of cataIytic surfaces and of molecules adsorbed on them is in a very early stage of development. It may, therefore, be useful to indicate a few directions in which further practical and theoretical advances may be expected. In extending the examination of catalysts,
112
R. H. QRIFFITEI
it is highly desirable that more attention should be given t o evaporated films of sulfides and oxides, which can be prepared by methods similar to those already used with metals. The determination of the structure of such films by electron diffraction and by microscopic examination, and the measurement of their surface areas, could then be related to their adsorptive capacities and catalytic activities under a variety of conditions. On the theoretical side also, importance must be attached to oxides and sulfides because these present many problems which do not arise with metallic catalysts. The functioh of the oxygen and sulfur atoms in such lattices is a matter of prime importance, and structuree produced by partial removal of these elements from their compounds must have very special properties. It would be of great interest to know, for example, in what positions the hydrogen atoms split off during ring-closure of aliphatic hydrocarbons are adsorbed on the catalyst; it seems highly probable that the oxygen atoms in the chromium oxide or molybdenum oxide are involved and that they contribute an essential part to the specific effects of these catalysts. There is a wide range of cracking and isomeration reactions involving many types of hydrocarbons whose mechanism is not understood, but which must be closely related to the structure of the catalyst surface. Owing to the complicated character of the catalyst used for some of these processes, and to the occurrence of side reactions leading t o fouling by heavy polymers, there are several practical difficulties to be overcome in solving such problems. As far as oxidation reactions are concerned, much less information is available. In many instances, as, for example, in the production of ethylene oxide by oxidation of ethylene with silver catalysts (Twigg, 39), the reaction takes place by collision with chemisorbed oxygen. It is, however, quite likely that instances will be found where the mechanism is entirely different and where two-point adsorption takes place. In most existing discussions on the geometrical factor, attention has becn focused exclusively on the demands of the carbon bonds, and those of the elements in the catalyst surface have been ignored. As long as the crystal lattice has an ionic structure, this is permissible, but it is quite clear that with other structures, such as that of molybdenite, this procedure is unsatisfactory. If the position of an ole& adsorbed by two carbon atoms is -C considered at right angles to the Mo plane of MoSz, the angle MoC would be 112' if Mo-C is 2.14A., Mo-Mo is 3.15A. and C-C is 1.54A. The C-Mo-Mo angle, therefore, should be 68". Now the S-M0-S angles in the crystal are either 82 or 136", of which the smaller is the one available above a bare plane of Mo atoms; the bond angle above the plane is 41', and the olefin must, therefore, be held in a tilted position. It should
GEOMETRICAL FACTOR IN CATALYSIS
113
be possible to relate an effect of this kind to the capacity of a known surface for adsorption of a hydrocarbon. These few suggestions will indicate that a better knowledge of the geometrical factors in catalysis can only be obtained by the combined results of several distinct methods of theoretical and practical treatment. One of the most important of these is to increase the information available about the nature of the bond between the adsorbed molecule and the catalyst, and about its dependence on the chemical properties of the catalyst elements. I n this article, no account has been included of the evidence provided by work on promoter actions or on catalyst poisons. Both these have made contributions to the study of the geometrical factor, and are likely to do so more extensively in the future, but these subjects are themselves of such importance in any report on catalysis that they should be the subject of a separate treatment. REFERENCES 1. 2.
3. 4. 5. 6.
7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.
19. 20. 21. 22. 23.
Burk, R. E., J. Phys. Chem. 30,1134 (1926). Balandm, A. A., Z . physik. Chem. 126,267 (1927);B2,289 (1928),and many later Papem. National Research Council, Twelfth Report of the Committee on Catalysis. Wiley, New York, 1940. Horiuchi, H., and Polanyi, M., Trans.Furaduy SOC.30, 1164 (1934). Farkas, A., and Farkaa, L., J. Am. Chem. SOC.60,22 (1938). a. Conn, G.K. T., and Twigg, G. H., Proc. Roy. SOC.London 171A,75 (1939). b. Twigg, G. H., Trans. Faraday SOC.36, 934 (1939). Twigg, G.H.,and Rideal, E. K., Trans. Faruday Soc. 36, 533 (1940). Farkas, A., Farkas, L., and Rideal, E. K., Proc. Roy. Soc. London 146A,630 (1934). Beeck, O.,Smith, A. E., and Wheeler, A,, Proc. Roy. SOC.London 177A,62 (1940). de Lange, J. J., and Visser, G. H., Zngenieur, Ulrecht 68,No. 26,24 (1946). Sheridan, J., J. Chem. SOC.1944,373;1046,133,301,305,470. Craxford, S. R., Tram. Foraday SOC.36,946 (1939,); Craxford, S. R., and Rideal, E.K., J. Chem. Soc. 1030 1604. Herington, E. F.G., Tram. Faraday Soc. 97, 361 (1941). Smith, D.F., Hawk, C. O., and Golden, P. L., J . Am. Chem. SOC.62,3221 (1930). British Intelligence Objectives Sub-committee, Final Report 447, His Majesty’s Stationery Office, London, 1946. Field Information Agency (Technical), Final Report 276 Joint Intelligence Objectives Agency, Waahington, D. C., 1945. Herington, E. F.G., and Rideal, E. K . , Proc. Roy. SOC.London 184A,434 (1945). Visser, G.H., Chem. Weekbtad 42, No.21,127 (1946). Turkevich, J., Fehrer, H., and Taylor, H. S., J. Am. Chem. Soc. 63, 1129 (1941). Grifith, R.H., and Morcom, A. R., unpublished dats. Biltz, W., 2.anorg. allgem. Chem. 186,373 (1930). Kazanskii, B. A., Plate, A. F., Balanova, T. F., Zelinskii, N. D., and Sergienko, S. R., Compt. rend. mud. sci. U.S. S.R. 27,658,664 (1940). Hummer, R. W.,and Taylor, H. S., J. Am. Chem. SOC.63,2801 (1941).
114 24. 25. 26. 27.
28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.
.
R. H. QRIFFITH
Herington, E. F. G., and Rideal, E. K., Proc. Roy. Soc. London 184A,447 (1045). Padoa, M., Rend. accad. nuz. lincei 16, 220 (1906). Starr, L).F.,and Hixon, R. M., J . A m . Chem. Soc. 66, 1595 (1934). Shuikin, N. I., Nikiforov, V., and Stolyarova, P. A., J . Gcn. Cheni. U.S.S.R. 6 , 279 (1936);7, 1501 (1937). Zelinskii, N. D.,and Yurev, Y. K., Rcr. 84B, 101 (1931). Moldavskii, B.L.,and Kumari, Z.I., J . Gen. Chem. U.S.S.R. 4,298 (1934). Cawley, C. M.,and Hall, C. C., J . Soc. Ch.em. Znd. 62, 116 (1943). Weatgren, A., 2. anorg. allgem. Chem. 239,82 (1938). Crawley, B.,and Griffith, It. H., J . Chem. Soc. 1938,717,720, 2034. Griffith, R.H.,Marsh, J. D. F., and Newling, W. B. S., publication pending. Yurev, Y. K.,Ber. 69B,410,1002 (1936);J . Gen. Chon. U.S.S.R. 7,1868 (1937). Pitzer, E.C., and Frazer, J. C. W., J . Phys. Chem. 46, 761 (1941). Alekseevskii, E.V.,and Frid, K. V., J . Gen. Chem. U.S.S.R. 16,3 (1945). Feitknecht, W.,and Mlirti, W., Helu. Chim. Acta, 28, 129,149 (1945). Eckcll, J., 2. Bleklrochern. 38, 918 (1932);39,423,807,855 (1933). Twigg, G. H.,Proc. Roy. Roc. London 188A,92 (1946).
The Fischer-Tropsch and Related Processes for Synthesis of Hydrocarbons by Hydrogenation of Carbon Monoxide* H. H. STORCH Chief, Synthetic Liquid Fuels Research and Development Branch, Bureau of Mines, Pittsburgh, Pa.
CONTENTS Page 115 11. Development of the Ruhrchemie Proceu . . . . . . . . 115 111. Development of Fischer-Tropsch Process Other Than the Ruhrchemie Proceas . . . . . . . . . . . . . . 123 IV. Processes Related to the Fischer-Tropsch Synthesis . . . . . 133 . . 138 V. Carbide Formation and Carbon Deposition on Iron Catalysts . VI. Kinetics and Reaction Mechanisms in the Fischcr-Tropsch Synthesis . 140 1. Pressure and Temperature Coefficients. Effect of Contact Timc. . . 140 2. Mechanism of the Synthesis . . . . . . . . . . 149 References . . . . . . . . . . . . . . . 153
I. Introduction .
.
.
.
.
.
.
.
.
.
.
.
.
.
I. INTRODUCTION Industrial and scientific development of the chemistry of carbon monoxide hydrogenation has increased in direct proportion to the growth of the art of contact catalysis. To such earlier processes as the water-gas shift, the reduction of carbon monoxide to methane, and the synthesis of methanol, recent developments have added the synthesis of normal and isoparafhs and Cp to Czo normal and branched-chain alcohols. Research on carbon monoxide hydrogenation at very high pressures has barely begun but has already yielded the synthesis of paraffin waxes of very high molecular weight which are not found in natural petroleum. The synthesis of polyhydric alcohols at very high pressures of hydrogen and carbon monoxide remains aa one of the prizes of further research.
11. DEVELOPMENT OF THE RUHRCHEMIE PROCESS In 1913, the Badische Anilin und Soda-Fabrik (1) disclosed that hydrogen and carbon monoxide would react at 100-200 atmospheres pressure in the presence of alkali-activated cobalt or osmium oxide supported on asbestos to form a liquid product containing alcohols, aldehydes, * Published by permhion of the Director, Bureau of Mines, U. S. Department of the Interior, Washington, D.C.
115
116
H. H. STORCH
ketones, fatty acids, and some saturated and unsaturated aliphatic hydrocarbons. This work led eventually to the development of the methanol synthesis by the same company. In 1923, Franz Fischer and Hans Tropsch (2), working in the Fuel Research Laboratories of the Kaiser Wilheltn Tnstit,iite,found that slknlized iron turnings catalyzed the production of nn oily liquid from hydrogen and carbon monoxide at 100-150 atmospheres pressure and 325425” C. This product was called synthol, and it contained alcohols, acids, aldehydes, ketones, esters, and hydrocarbons. Fischer and Tropsch found that, at lower pressures of about 7 atmospheres, the oil obtained was almost completely paraffinic and contained a small fraction only of oxygenated organic compounds. At the lower pressure, however, only a very low yield of oil was obtained. In 1925-1926, Fischer and Tropsch (3, 4) developed much more active catalysts for the synthesis; these consisted of promoted cobalt, iron, or nickel. Use of these catalysts provided a fair yield of oil even at 1 atmosphere pressure, with only minor amounts of oxygenated organic compounds in the product, which was labeled “kogasin. Experimental work with Co-MnO-Cu and other catalysts was reported by Elvins and Nash (5, 6) in England in 1926-1928 and by Smith, Davis, and Reynolds (7) of the United States Bureau of Mines. In 1928-1929, Fischer, Tropsch, and Koch (8, 9, 10) published the results of an analysis of the oil product from an Fe-Cu catalyst operated at atmospheric pressure and 250°C. The fraction boiling in the range 60-185°C. was 70% olefinic. The 30% saturates contained octane, nonane, and isonane; no diolefins or naphthenes were found. Minute amounts (0.144%) of beneol and toluol were present. The isononane was probably %methyloctane and was present in very small amount. A very small amount of 3,3-dimethylpcntene-l also was present. In 1928, an analysis was published by Smith, Hawk, and Reynolds (11) of the product from a Co-Cu-MnO catalyst. This product, although more saturated, was similar to that obtained by Fischer and his coworkers from an Fe-Cu catalyst. In 1930-1932, Kodama and other Japanese experimenters (12, 13, 14, 15) published data on the efficiency of difierent promoters such as Cu, Thoz, WOa, Moos, MgO for the cobalt catalyst. In 1930, Smith, Hawk, and Golden (16) discovered that ethylene, when added to the synthesis gas, enters into the oil-producing reaction on a Co-Cu-MnO catalyst but not on an F e C u catalyst. In the latter case, most of the ethylene appears as ethane in the products. It was pointed out also that water was the main oxygenated product from the cobalt catalyst, whereas carbon dioxide was the chief corresponding product on the iron catalyst. The behavior of ethylene when added to synthesis gas and passed over a cobalt catalyst was verified by Craxford (17) in 1939.
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
117
Storch (18) reported data on C1to Caratios in the end gases from the synthesis as follows: 1O:l on promoted cobalt, 4.5:l on unpromoted iron, and 1 :2 on iron-copper catalysts. Any postulated mechanism of the synthesis should provide a satisfactory explanation of these facts concerning the function of ethylene in the reaction. In 1930-1932, Fischer and his coworkers (19, 20, 21) developed more active promoted nickel and cobalt catalysts, and increased the yield per pass to over 100 g. oil/cubic meter of 2 H,+1 CO gas. I n 1932, Fischer and his coworkers (22, 23, 24, 25, 26) started process development work, using a Ni-Mn0-A1203-kieselghr catalyst. Because of the highly exothermic character of the reaction, he found it essential that the path of. heat transfer from a granule of catalyst to a cooling surface be less than 10 mm., preferably about 7 mm. His first experimental plant consisted of a battery of cells 12-mm. thick by 1.2-meters wide by 5-meters high, immersed in an oil bath. A yield of 70 g. oil/cubic meter of 2 Hz+l CO was obtained a t 190-210" C. and a t atmospheric pressure. In 1933, the Ruhrchemie A.G. (27,28,29,30,31) erected a pilot plant of about 1000 tonsjyear capacity at Oberhausen-Holten. A Ni-Mn0-A12O3-kieselguhr catalyst was used a t 19Ck21O"C. and at atmospheric pressure. The difficulties encountered (32) were a short catalyst life of 4-6 weeks, despite an elaborate and expensive sulfur-removal process, expensive design of converters, loss of catalyst metals and promoters during catalyst regeneration, and conversion of kogasin to marketable products. In the period 1933-1939, an intensive and rapid development of more active catalysts occurred, with final selection of Co-Th02-Mg0-kieselguhr (100:5:8:200 parts by weight) as the best catalyst for industrial operation (33, 34, 35, 36). Table I (37) contains the results of laboratory tests at 1 atmosphere and at 7 atmospheres pressure, using this catalyst. Koch and Hilberath in 1941-1942 (38, 39) studied the composition of the fraction of the product from the cobalt catalysts which boiled below 100" C. The saturated C4 fraction contained 7.5% isobutane. The remainder of the product fraction boiling below 100" C. contained, in addition tonCa, ~ C E and , nC7,also methylbutane, 2-methylpentane,3-methylpentane1 %methylhexme, and 3-methylhexane. Only traces of more highly branched paraffins were found. In 1939, von Weber (40) had estimated the degree of branching of different fractions of the product from measurements of the boiling points and molecular weights of fract,ions of Fischer-Tropsch gasoline and reported: Fraction Molfraction of branchcd hydrocarbons
CS
0.15
C, 0.20
CS
0.27
co
0.35
ClO 0.40
These data show that 1 out of every 25 to 50 carbon atoms of the Ceto Clo
118
H. H. STORCH
fractions is a tertiary carbon. No quaternary carbon was found in either von Weber's or Koch and Hilberath's work. In the Bureau of Mines, a recent infrared spectrometric analysis of the Ce fraction from a cobalt catalyst showed 10% 2-methylpentane and 5% 3-methylpentnne. TABLE I Product from 100 C0:6 'l'h0,:T.b hfgO:800 Kieselguhr Catalyst at 176-200° c. and at ( A ) Almsph&, ( B ) 7 Atmospheres Pressure of 2Hz 1CO Gas, 1840% Inerta. N o Recycle.
+
Weight
% . of
kquids ldealgae,
Fraction
26 24 70 77 41 34 Hard wax
plya
-A B A B g./m.'
21 17 51 58 51 61
4 43
solids
.. ..
..
56 62 33 27
35 40 35 35
Denait MO*&
A
-
Olefins vo~urn;
B
CFR
Pour p+t,
% A B
A
..
..
c.
.. .. 5030 . . 0.689 0.685 37 20 . . .693 .689 34 18 . . .780 .760 15 10 -18 .766 .766 13 8 -9
octane
B
___ number A g-
Cetene number
2
-
B
..
..
. . . .
. .
..
. . . .
. .
52 28 49 25
. . . .
. . -7 -2
. . . . . . . . 100 100
. . . . 100 105 . . . . . . . . . . . . . . . .
3 30 J" .930 ~~~
.~
-
..~-
a Higher peroentaur of wax may bo o b t a i n 4 from Co-Mncatalpts at 166-1fl8" C. and LO atmospheres,
but with no increase in Diesel oil yield.
All of the German Fischer-Tropsch plants in 1938-1944 ( 4 1 4 9 ) were operating according to the Ruhrchemie process using the Co-ThOz-MgOkieselguhr catalyst at 180-200° C., at either 1 or 10-atmospheres pressure, and with 2 or 3 stages with product recovery after each stage. The average yield was 150 g./cubic meter of 2Hz+lCO gas of hydrocarbons ranging from propane-propylene to waxes of 2000 molecular weight. The space velocity employed was 60 to 100 volumes of feed gas/volume catalyst space/hour. The normal synthesis gas required for the Fischer-Tropsch process is a mixture of 2 volumes of hydrogen and 1 volume of carbon monoxide. Recent practice in the Ruhr when using cobalt catalyst showed ratios of from 1.8 to 2.0 for the atmospheric pressure process and ratios as low as 1.5 for the medium-pressure process. In the Ruhrchemie plant at Sterkrade, the medium-pressure process used different ratios for each of the three stages of the synthesis, 1.4, 1.6, and 1.8, respectively, by introducing the requisite amount of "converted" water gas before each stage. In all cases, the synthesis gas contained inert constituents which seldom exceeded 20% by volume.
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
119
In all plants in the Ruhr, synthesis gas was manufactured from coke in standard design “blue-water” gas generators. In a few plants, this process was supplemented by other processes such as cracking of coke-oven gas. In order to obtain the higher ratio of hydrogen to carbon monoxide required for the standard Fischer-Tropsch operation, it was necessary to pass a portion of the water gas, mixed with an excess of steam, through converters containing a “shift” catalyst whereby carbon monoxide was converted into hydrogen and carbon dioxide. An iron oxide-chromium oxide catalyst was used in this step at 450-500” C. In two plants in the Ruhr, a high-hydrogen gas was produced by cracking or re-forming cokeoven gas, and the product was mixed with water gas, thereby reducing the volume of water gas to be converted. At the Krupp Fischer-Tropsch plant a t Wanne-Eickel there was a large commercial installation of the Krupp-Lurgi process for manufacture of low-temperature coke. Most of this coke was gasified in blue-water-gas machines and yielded a hydrogen to carbon monoxide ratio of 1.35,which is closer to the desired ratio for synthesis than is obtained from high-temperature coke. Several processes for the complete gasification of coal were developed in Germany, and, although they were used for production of hydrogen for coal hydrogenation or synthetic ammonia production, the same plants, with only minor modifications, could have made synthesis gas. The Pintsch-Hillebrand, the Schmalfeldt-Wintershall, and the Winkler were three such processes. The Fischer-Tropsch process requires synthesis gas in which the total sulfur does not exceed 2.0 mg./cubic meter. This purification was done in two stages: (1) Removal of hydrogen sulfide and (2) removal of organic sulfur. The removal of hydrogen sulfide is almost universally carried out by the well known iron oxide process. At the Luetzkendorf plant, the socalled “Alkazid” process had been installed, in which a solution of an alkaline organic compound absorbs the hydrogen sulfide, which is then continuously stripped from the solution by steam. At Luetzkendorf, the evolved hydrogen sulfide was converted into elemental sulfur. Organic sulfur was removed in all of the German Fischer-Tropsch plants by passing the synthesis gas through an alkalized iron oxide catalyst (containing about 30% sodium carbonate) at 280” C. The reaction is essentially catalytic oxidation of the organic sulfur in the gas, and it is necessary that a small percentage of oxygen be present in the synthesis gas. The details of preparation and reduction of the Co-ThOTMgOkieselguhr catalyst are described in CIOS reports (41-49)and other publications (34,35, 36). A solution of cobalt and thorium nitrates is mixed with sodium carbonate solution so as to completely precipitate the cobalt and thorium carbonates. A slurry of magnesium oxide and kieselguhr is
H. H. BTORCR
120
mixed with the suspension of cobalt and thorium carbonates, the mixture rapidly filtered, washed exhaustively, and the filter cake dried. It is then crushed and screened to obtain 1- to 3-mm. granules. These are reduced a t 350-450" C. in a stream of dry carbon dioxide-free hydrogen flowing at a space velocity of 3000 to 4000 volumes/volume of catalyst. The reduced granules are stored in an atmosphere of carbon dioxide in steel containers used to ship the catalyst to the synthesis plants. The converters in atmospheric pressure Fischer-Tropsch plants consist of rectangular steel boxes about 15-feet long, &feet high, and &feet deep. These contain vertically placed steel sheets about 2-mm. thick, spaced about 7 mm. apart. The spaces between the steel plates contain the catalyst. The steeI sheets are perforated by horizontally placed cooling pipes about 5-cm. o.d., through which water under pressure is circulated. The cooling tubes are manifolded and connected to a steam boiler which generates steam for power and heating purposes. The converters in the mcdium-pressure (10-15 atmospheres) Fischer-Tropsch plants are essentially similar to tube boilers, except that the tubes are double, with the catalyst contained in a narrow annulus between two concentric tubes. A top and bottom weld through the outer tube connects the inner tube with the boiler space, so that it is filled with cooling water. The annular layer of catalyst haa an inside diameter of 24 mm. and an outside diameter of 44 mm. and is 15-feet high. Each converter contains about 2000 such tubes welded into tube sheets and is about 8-feet in diameter. , The weight, volume, cooling surface, and capacity of the atmospheric- and mediumpressure converters are listed in Table 11. Table I11 shows the number of converters in different stages a t several of the German plants. TABLE I1 Ruhrchemie Converter Characlerislics _ _ ~
_._______ _______. .~~____ ____
__
Characteridtic
__ _Wt:ight (tons) Volume (total) cu. ft. Volume (cntnlyst) cu. ft. Finned surface sq. ft. Tube surface sq. ft. Capacity barrels of primary product per day
Atmospheric prcsaure
45 650
350 38700 4300 18
,
~~-~ ___ ___ ___ Middle preasure
45 950 350
None 22600
18
Rheinpreussen at hloers operated 96 conventional Ruhrchemie atmospheric pressure reactors, of which 60 were normally in the first stage and 30 in the second stage. The operating temperature was in the range 195-
121
PROCESSES FOR SYNTHESIS O F HYDROCARBONS
200” C. The synthesis-gas charging rate averaged about 630 cubic meters/ hour,/oven. The total liquid product, including “GRsoI” (C, and C,), TABLE I11 Multi-Stage Operation at Several German Plants
_ _ ~ _ _ _
Number of converters Name
Location
Stage 1
Sage
Stage
2
3
16-20 18-20 48 Total 30 63 Total
4-8 14
Hoesch-Benzin Ruhrbcnzin
Dortmund Stcrkrade
40 36
Rlieinprcussen Klockncrwerke
Rloem
Cnstrop-Rauxel
60
0
Prewure. atm.
10-15 11 1 1 1
averaged 15&155 g./cubic meter. The product distribution is shown in Table IV. TABLE
IV
Rheinpreussen Product Distribution Constituent
Per cent
GMOI(C,+CJ Gasoline Diesel oil Soft paraffin wax Hard paraffin WRX
10.0 52.2 26.5 7.6 3.7
Ruhrbenzin at Sterkrade-Hol ten operated about 48 atmosphericpressure reactors and 72 middle-pressure reactors, all of conventional design. The normal pressure ovens are operated in two stages on charge gas containing the usual 2 volumes of hydrogen and 1 volume of carbon monoxide. The arrangement of the middle-pressure reactors is shown in Fig. 1 (42). In this operation, the Hz to CO ratio of the inlet gas to the three stages is adjusted to 1.4, 1.6, and 1.8, respectively, by introducing the requisite amount of converted water gas before each stage. For Stage 1 of the medium-pressure operation, the inlet gas rate is 1000 cubic meters/hour/oven, the temperature 180-200” C., and the gas contraction 50%. Similar conditions obtained for Stage 2. I n Stage 3, where all newly charged ovens are started, higher rates, namely, 7000 cubic meters/hour/oven, and lower temperatures, 165-185’ C., can be employed. The over-all contraction is about 75% and the yield, 150 g. Csand higher hydrocarbons per cubic meter of ideal gas (that is, 2H2+1CO). The
122
H. H. STORCH
advantage of starting with a gas relatively deficient in hydrogen and then increasing the proportion of hydrogen in later stages lies in combining the beneficial effects of low hydrogen partial pressure on olefin content, and
18-20 wins -H&1.1.6:1
Dlmt condmsrr
Fro. 1. Flow scheme for synthesis cycle used at Sterkrrtde. (Riihrchcrnie,A.G.-Ruhrbenzin A.G., Stcrkrade), Finuren I - 1 1 pubbhed by oourtesy of the Bureau of Miaen. U. 8. Department of the Interior.
methane production with the efficient utilization of the carbon monoxide achieved with normal synthesis gas (2Ha:1CO). The products produced by this method of operation are shown in Table V. TABLE V Product at Ruhrbenzin, Sterkrade-Holten __ - . Weight. 9% total product
Produrt
GwoI (C,+CJ C , to 170"C.= 170-280O C. 280-340"C. (soft wax) Raidue (hard wax) _________ - -~
~-
10 25 30 20 15
Per cent of oleflna in fraction
40 24 9
..
..
Motor octane number, about 45.
When normal synthesis gas (2H2+lC0) is used throughout, 14% of the total hydrocarbons produced is met,hane, but with the above scheme, the proportion is reduced to 10%. The middle-pressure synthesis yields 2 to 3 tjmcs as much wax as the synthcsis a t atmospheric pressure.
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
123
111. DEVELOPMENT OF FISCHER-TROPGCH PROCESSES OTHER THAN THE RUHRCHEMIE PROCESS All of the commercial German Fischer-Tropsch plants in 1938-1944 were operating the Ruhrchemie process, using a cobalt-thoria-magnesiakieselguhr catalyst a t about 200" C. either at 1 or 10 atmospheres pressure, and with two or three stages with product recovery after each stage. The only outstanding process development using the cobalt catalyst consisted in recycling about 3 volumes of end gas from the first stage per volume of fresh gas with product condensation after each cycle (41-53). On a pilotplant scale, this development resulted in an increase of 30% in the throughput without sacrifice in yield per cubic meter of 2 H,+1 CO gas, or in an increase of the yield from 150 to 170 g./cubic meter a t the same throughput. Using the recycle process and charging water gas (1 Ha+l CO) instead of 2 Ha+l CO, the olefin content was increased from 20 to 55% in the liquid fractions of the product. When water gas is used in the older process without recycling, a low catalyst life results because of carbon deposition. The recycle gases lower the partial pressure of the carbon monoxide sufficiently to avoid excessive carbon formation. The recycle-process gasoline has a 50-55 motor octane number as compared with 45 for the older process without recycle. Refining the gasoline from the recycle process by passage over activated clay increases the octane number to 70 without appreciable decrease in yield. The cetane number of the fraction boiling from 120 to 280" C. was above 100. The Lurgi and I. G. Farben companies were active during 1938-1944 in process development using iron catalysts (50,53). I. G. Farben operated a pilot plant using a sintered iron catalyst and a hot-gas recycle process. The catalyst was prepared by making a paste of iron powder (obtained by decomposition of iron carbonyl) with a dilute solution of sodium borate. The extruded and dried granules were heated in a hydrogen atmosphere at 850" C. for several hours before use in the synthesis, which was conducted in the range 320-330" C., 20 atmospheres, and %-second contact time. The gases from the converter were recycled through a heat exchanger to remove heat of reaction. Conversion per pass was about 1%, so that about 100 volumes of recycle gas to 1 of fresh gas entered the converter. The gas is passed over the catalyst so rapidly that the temperature does not rise beyond the optimum upper limit (about 350°C.). The heat evolved in the synthesis (about 4000 kcal./kg. of product or 7200 B.t.u./ pound) is absorbed in an external heat exchanger, using a 10"C. drop in temperature of the recycle gas. The conversion is about l%/pass; therefore, 1% of fresh synthesis gas is added per cycle, and an equivalent amount of gas from the reactor is removed and passed through a condensation and product recovery system and thence to a second stage, which is about
II. H. STORCH
124
one-quarter the size of the first stage. In between the two stages, the gases are cooled to 0" C. and carbon dioxide is scrubbed out. All of the available yield data are ,extrapolations of results obtained in a single stage in a pilot plant operated for several months using 400 1. of catalyst, and a hydrogen to carbon monoxide ratio of about 0.9. The conversion in the first stage was 78a/,, and the conversion in a second stage was estimated at 13%%. The yield per cubic meter of fresh gas was about 160 g. hydrocarbons, of higher molecular weight than ethane. The space-time yield was about 1 kg. of product/l. of catalyst/day (this is about eight times that from the Ruhrchemie cobalt catalyst process). The products were reported t o contain : 1. 44% unrefined highly olefinic gasoline of 75-81 research octane number. Upon refining, a 90% yield of 84-86 research octane number gasoline was obtained. 2. 11%Diesel oil of 47-53 cetene number and -25' C.pour point. 3. 7y0 Cb and 8% CI. (The CBand C4were 80% unsaturated.) 4. 5% CzH4. 5. 17% CHa+CaH8. 6. 7% alcohols, chiefly ethanol and propnnol. 7. 1%paraffin wax.
This report is remarkable in that only 17% methane plus ethane is specified. Others who have experimented with iron catalysts a t 300-320°C. , with little or no recycle of end gases, have reported much higher yields of methane ar:d ethane. The cost of operation of this process is markedly dependent upon the pressure drop across the catalyst bed. This was reported to be 1 meter water/meter of catalyst depth. Increased gradients would result with increasing carbon deposition, and the operating temperature is dangerously close to that a t which a rapid rate of carbon formation occurs. Other processes in the development stage by I. G. Farben included a liquid-phase operation in which iron powder prepared from iron carbonyl is mixed with a high-boiling oil and synthesis gas bubbled through the mixture. This process yields more Diesel oil (of 6&70 cetene number) and much less methane and ethane than the hot-gas recycle process. The gasoline from the liquid-phase process operated at about 300" C. is reported to have a research octane number of 90. Another I. G. Farben development was an oil-recycle process in which a cooling oil was passed concurrent with the synthesis gas over granules of an iron catalyst (synthetic ammonia type, Fe304 2.5~0A1z08 O.5%K2O). The cooling was effected by recycling the heated oil through an external heat exchanger. Development work by the Lurgi Company (21) involved n recycle
+
+
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
125
operation using a Ruhrchemie middle-pressure (10-20 atmospheres) converter. This recycle process, using cobalt catalysts, has been described above. In the same pilot-plant unit used for the recycle process work on a cobalt catalyst, the Lurgi Company developed similar processes using iron catalysts. This unit, which was installed at Hoesch-Benzin at Dortmund in 1938, consisted of a single tube of the standard middle-pressure reaction as built by Gutenhoffnungshutte. The annular layer of catalyst had an inside diameter of 24 mm., an outside diameter of 44 mm., and a depth of 5 meters. The tube was water-jacketed with a vapor chamber connected to the top by means of which the steam pressure could be controlled to give the desired synthesis temperature. Dr. Herbert (Research Director of Lurgi) stated that the best iron catalyst developed by his laboratory has the following composition : 100 Fe, 25 Cu, 9 A1203,2 KzO,30 Si02. The catalyst is prepared by dissolving copper and aluminum nitrates in a 10% solution of ferric nitrate in such quantites as will give the specified ratio of metals. The solution is heated to boiling, and a 10% solution of sodium carbonate is added rapidly a t about 70°C. in the quantity required to precipitate the metals as hydroxides. The kieselguhr is then added rapidly, stirred for about 1 minute, and the mixture washed to a pH of 8.0, after which it is washed with a potassium carbonate solution to incorporate the specified quantity of KzO. The product is dried in a centrifuge sufficiently to permit extrusion, and is dried further on a conveyor belt by a blast of hot air to facilitate cutting into desired lengths. Final drying is done at 100" C. This catalyst is preferably reduced in the synthesis unit with hydrogen at 250-350" C. for 1 4 hours. The rate of hydrogen flow is 1cubic meter/lcg. catalyst/hour. Reduction with synthesis gas is possible. Synthesis is started at about 180" C., and the temperature is raised to 220" C. in 2 days. The operating pressure was 20 atmospheres. The space velocity (volumes of gas per volume of catalyst per hour) of fresh gas is 100, of recycle gas 300, and the tail-gas volume is 48% of the fresh-gas volume. Over an operating period of 3 months, the temperature is raised from 220 to 230" C. This catalyst has not been run to exhaustion by Lurgi, but it is believed that its life would be about 1 year. Other conditions of the operation and yields are shown in Table VI. An iron catalyst which yields a higher proportion of gasoline in the product is made by impregnating Luxmasse with a solution of copper ammonium nitrate to obtain 3% by weight of metallic copper in the finished catalyst. The resulting catalyst is not highly active and requires a high operating temperature, but at this temperature it gives a relatively low boiling product, with less catalyst deterioration due to carbon depositiop than would occur with a more active catalyst. The operating temperature
126
€1. Ii. STORCII
is 275" C. and pressure 20 atmosphcrcs. The space velocity of fresh gas is 100, of recycle gas 300, and the tail gas volume is 50.5% of the fresh gas TABLE VI LtcTgi Rc.c?lcle Operation for H i g h W a x Produclion .. _.
~-
~~
~
GAS COMPOSITION
cox
:
Fresh gas Total fccd gas
CO 37.6 269 22.6
5.8 229 28.2
Tail gas
48.1 31.8 27.0
CIT, 0.1 2.0 2.7
N2
CnHm
8.4 7 5.0 17.6
0.0 1.4 1.9
YIEI,nS, G./NORMAT. CURTC METER FROM TWO-STAGE OPlTRATION:
60 22 29" 29h
Hard w i n Paraffin
Oil Gnsoliric
Alcohols Cnrol (C,+C,) C,+C*
9 15' 22
f Olefins, 4e0;.
Oletins. (id'%.
' 0leE.u GOYp.
Total yield of Iiiiuid productn plus g w u l equals 170 u..'nornral cubic meter.
volume. Wax, oil, arid gasoline are removed before recycling. The gas composition and yield and product distribution are shown in Table VII. TABLE VII
Lurgi Recycle Operation for High Gaaoline Production GAS COMPOBITION:
CO,
co
H2
CH,
Fresh gas Total feed gns Tnil gas
2.8 31.8 48.0
53.2 29.8 16.7
35.6 19.8 11.3
0.1 3.5 5.3
Nn
CnHm
8.3 14.0 16.5
1.3 2.2
,
..
YlELD G./NORMAL CUBIC METER FROM SINGLE-STAGE OPERATION:
Pnrnffin Oil Gasoline
Alcohols Gnsol (C,+C,) G+C*
8 20"
70b ~~~
.
5
32" 35
-.
~~
f Olefin content. 60%.
Olefin contnnt 75%.
Olefin content' 70%. Total yield of liquid producta plus
gas01 equals
135 g./normal cubio meter.
In pilot plants identical with the Liirgi Company unit of the middlepressure converter described above, simultaneous tests were run on six different iron catalysts (54). Tables VIII, IX, and X contain the data on these comparative tests, conducted a t the Braunkohlc-Renzin A.C. in Schwartzheide-Ruhland. One catalyst mas submitted by each of the following organizations: Kaiser
Y
d
0
TABLE VIII
d M ul ul
Calalysts Used in Coopcrdive Tests d Schwartzheide
Company
laiser Wilhelm urgi rabag
C . Farbenindustrieo .uhrchemie .heinpreussen
Copper (me)
KEOI
Carrier
Parts/100 parts of iron
1 10 10
5
a Contained 2% (rU101+CaO).
gemit, g.h e .
0.75 9 K20 from K&iO, 0.5
None 24 SiOzfrom K S O , None
1.02 0.79 1.37
1 0.5 to 2 0.5 to 1
None 50 Kieselguhr Powdered dolomite
2.27 0.44 0.68
(10)
.. 5
A parent
2! Reduction
(Induction)
2H?+1 C0,325"C. Hzr30% Reduction Rater gas, 245"C. (Hz or 2 Hz+1 CO, 225" C.) H2,500"C. Complete reduction Hf Hl, 300-400"C. (Watergay, 245"C.)
kl
0
z rn
B
B
0 4
a
3
z
0 d P
0
m
H. H. BTORCH
128
Wilhelm Institute, Lurgi, Brabag, I, G. Farbenindustrie, Ruhrchemie, and Rheinpreussen. Because the objective was to rcplace cobalt catalysts in existing plant, the highest permissible temperature in these tests was 225' C., and the TABLE IX Cooperative Tcsls on Iron Catnlysls at Sch.umrtzheide (10 Atmospheres, 200-225" C. synthesis gas wm 88% 1 COf1.25IT2 nnd 12% iwrts. Yields per cubic mctcr are bused on feed g&9. Space velocity L volumrs (N.T.P.) of feed gas pcr volumc of catalyst per hour.) .-
~~~~
K&er Wilhelm Institute
Lurgi
Brabnr!
I. .G. Farbeniridustrie
Ruhrchernie
Rheinpreusaen
109 85 0.80
107 88 0.66
111 77 0.69
115 81 0.74
104 70 0.72
104 62 1.07
147
142
141
144
147
168
125
124
108
117
103
I04
~~
Space velocity CO convcrtcd, % CO:Ho, ucle ratio Maximum yield, g./cubic meter Avcrage yicltf, g./cubic meter Avcrage yield, tona/day Kg.Cs+/hr./m.'
3.26 10.9
2.88 10.8
3.19 11.6
3.23 10.9
2.57 8.8
2.60 8.5
TABLE X Conperdive Tests on Zron Catalysts at Schwartzheide Produd Distribution in Weight Per Cent K&er Wilhelrn Institute
Wax (Olehs) Diesel, CIIto C,, (Ol(!finS) Grtsolinc, cs to c,o
c,+c4 (Oletins) CJ% CaH 8
CHd
CI
+c,+C* (AlcohoLs)
18.1 (2) 14.2 (15) 25.0 (48) 19.8 (531 0.9
7.8 8.0
6.1
.~ ~~
Ruhrchemie
Rheinpreuasen
25.6 12.8 (23) 22.2 (44) 14.5 (62) 2.2 4.4 7.0
13.1 (2) 13.5 (1 9) 29.6
10.6 (76) 2.2 2.3 3.8
21.0 (12) 11.2 (45) 25.4 (59) 18.6 (71) 2.8 4.9 9.0
4.2
7.1
11.3
Lurgi
Brahag
35.2 (13) 13.2 (39) 19.4 (51) 12.2 166) 2.7 2.6 5.6
43.0 (11) 16.3 (47) 17.6
(67)
9.1
-
I. G. Farbenindustrie
_____ __ ___. ~
(Figurea in parenthws are per oont of olefins in thi, fraction.)
..
____-
(8)
(48)
21.5 (53) 0.7 8.1 8.1 5.4
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
129
operating pressure was 10 atmospheres of 1.25HZfl CO gas (containing 12y0 inert gas). The analysis of the synthesis gas showed 6.2a/, COa, 39.2qib CO, 48.8% H2, 2.60/, CHI, and 3.2% Nz. The duration of the tests was 3 months. The temperature and space velocity were independently varied by each operator to secure optimum yields and durability. At the end of 2 weeks all units had reached about 200" C. The space velocities were in the range 105-110 volumes of synthesis gas per volume of catalyst per hour. Only three of the six catalysts operated 90 days with one filling of catalyst. The others encountered "coking" difficulties and had to renew the charges repeatedly (four times in the case of Rheinpreussen). A description of the method of preparation of only one of the catalysts was found. The catalyst submitted by the Kaiser Wilhelm Institute was prepared in the following way: The iron was precipitated from the boiling dilute nitrate solution by the boiling soda solution. One per cent copper (based on the iron) had been added to the iron solution before precipitation. The precipitate wm filtered and washed with hot distilled water until free from alkali. A paste was then made with distilled water and a solution of potassium carbonate (Containing 1% of potassium carbonate, based on the iron) was mixed with the paste. The paste was thickened on a waterbath and drying completed at 105' C. in an oven. The dried pieces were then broken into granules of 2 4 mm. The pretreatment of the catalyst was as follows: A mixed gas, rich in hydrogen (2Ha+lCO), was passed over the catalyst for 24 hours a t a pressure of 1/10 atmosphere and a temperature of 325" C., (8 1. gas/hour/ 10 g. iron). The strongly pyrophoric catalyst was then soaked in paraffin wax to protect it from oxidation, and put into the converter with as little contact with air as possible. At the beginning of the synthesis, the catalyst was very active at 185190" C. To maintain a high conversion throughout the 3-month test, the temperature was raised gradually from 195 to 224" C., most of the operation being conducted between 215 and 220" C. The Ruhrchemie Fe :Cu :CaO :kieselguhr catalyst was probably prepared as follows: Calculated amounts of iron and copper turnings and calcium carbonate are dissolved in nitric acid, to yield a solution containing, in grams per liter, 115-125 g. iron, 5-7 g. copper, 11-13 g. calcium oxide, and no excess of nitric acid. The iron is in the ferrous state. Before precipitation, this solution is diluted with distilled water to lower the iron concentration to 3 6 5 5 g./L This solution, heated to 98" C., is run into a boiling solution of 90-100 g. soda ash/l., while being stirred vigorously. The quantity of soda ash used is such that, at the end of the precipitation, the pH is 6.8. Then 30 parts kieselguhr,'100 parts iron are added to the mix-
130
H. H. GTORCH.
ture. The total time of precipitation must not exceed 5 minutes. The mixture is filtered rapidly and washed with hot water until the filter cake contains 0.5% of sodium nitrate based on its content of iron. More complete washing is undesirable because of loss of calcium carbonate. About 200-220 cubic meters of wash water per ton of iron is sufficient. After washing, the filter cake is repulped and enough potassium hydroxide added so that after a subsequent additional filtration 3.0-3.5 g. KOH/100 g. iron are retained. This is achieved if the concentration of KOH in the mother liquor is about 6 g/l. After impregnation, the suspension was filtered and the filter cake dried at 110" C. and subsequently crushed and screened. This procedure yields a catalyst containing Fe:Cu:CaO:kieselguhr in the proportions 10:5:10:30 parts by weight. The I. G. Farbenindustric catalyst was fused Fe304 containing 2% AlaOa+CaO and 1%KZCO3. The Lurgi, Brabag, Ruhrchemie, and Rheinpreussen catalysts probably were precipitated from nitrate solutions. The presence of about 35% of calcium carbonate in the Rheinpreussen catalyst may have been responsible for the relatively rapid formation of carbon. The silica and kieselguhr in the Lurgi and Ruhrchemie catalysts also are known to decrease the durability of iron catalysts. The CO to Hs consumption ratios given in Table IX show that 50% or more of the oxygen appeared as water. This is much greater than the yield of water reported in laboratory tests by the Bureau of Mines and the Kaiser Wilhelm Institute. The yields of liquid products obtained from the Kaiser Wilhelm Institute and the Lurgi catalysts were only about 20% below those recorded for cobalt catalysts a t 10 atmospheres. The highest wax yields (see Table X and Fig. 2) were obtained from the Lurgi and Brabag catalysts, whereas the Rheinpreussen catalyst had the lowest wax yield. The Diesel oil yield was lower than the gasoline yield for all of the oatalysts. Despite a higher yield in grams per cubic meter of synthesis gae per paas through the converter, the space-time yields were about the same aa those obtained in Bureau of Mines laboratory tests, using unpromoted iron catalysts precipitated from ferric nitrate solution with potassium carbonate (55). In the United States, several oil companies have developed a fluidized iron catalyst process for producing motor gaaoline by the Fischer-Tropsch process (56-64). In the pilot-plant work on this fluidized-iron process, doubly promoted synthetic ammonia catalyst, Fea04,A l a 0 3 , and KZO, 97, 2.5, and 0.5%, respectively, has been used. The fused catalyst is finely ground and completely reduced by hydrogen before use. It is then introduced while protected by an inert atmosphere, such as COI or oxygen-free flue gas, into the converter. The pilot-plant converters are about 12-inches i.d. by 24-feet high, and oontain an internal heat-exchange surface. This
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
131
heat exchanger consists of a bundle of about 12 1-inch closed-end tubes, each containing an inner .tube through which water or Dowtherrn is circulated at the pressure corresponding to the operating temperature. In an alternative design, the converter is jacketed and the internal heat-exchange tubes are connected to this jacket at the top and bottom of the converter. I00
90
I
40
3 30 20
I
PRODUCT
I G. DISTRIBUTION
I
RCh
FIG.2. Co-operative tests on iron catalysta at Schwartzheide. = =Olefins, OH =Alcohols, OR=Esters, K.W.I. =Kaiser Wilhelrn Instit,ute, L =Lurgi, B = Brabag, I.G. = I.G.Farben, Rch = Ruhrchemie, Rhp = Rheinpreussen.
Pilot-plant operation is conducted usually at 300-325’ C., 20-atmospheres pressure, with recycle of 3 4 volumes end gas/volume fresh feed. The latter is 1.8 to 2.0 Ha to 1CO, prepared by oxidationofnaturalgaswith oxygen-steam mixtures at 300400 p.s.i. and 1200” C. The synthesis operation is in “fixed” fluidized bed; that is, the bed of catalyst is suspended in the flowing gas with no “carry-over” of powdered catalyst outside the converter. Carry-over is completely avoided by the use of Aloxite filters in the expanded section of the top of the converter. Catalyst density in the fluidized bed is 60-80 pounds/cubic foot at the start. This density decreases during the first 2-14 days of operation to 10-20 pounds/cubic foot because of “spalling” of the catalyst induced by carbon formation. With a hydrogen-rich total feed gas, 1600 hours of continuous operation waB achieved, and with a carbon dioxide-rich feed gas, only about 400 hours. Operation is limited to 1.5-2.5 linear feet/second gas velocity; at this velocity conversions of 90-95% are obtained per pass so that multistage opera-
132
H. H. STORCH
tion is unnecessary. Carbon dioxiderich feed gas operation results in a product which is chiefly highly olefinic gasoline with minor amounts of Diesel oil and oxygenated compounds (chiefly C&S alcohols, plus some aldehydes and ketones). Only L fraction of 1%wax is produced, but this is sufficient to occasionally cause defluidiaation of the bed. The gasoline, with only little refining, is 1~ suitable motor fuel of about 75 motor octane number which can be appreciably increased by addition of polygasoline from the Coand CSolefins and by tetraethyl lead. Based on this pilot-plant work, a plant of about 8000 barrels/day has been designed and is to be erected at Brownsville, Texas. Synthesis gaa is to be made by reacting natural gas at about 400 p.8.i. with oxygen and TABLE XI Space-Time Yield of Variou Fixher-Trope& Ptoceaeea --___ Liquid produot dmtributlon Catalyst
Temp., "C.
co co
175-200 175-200 200-225
__
Pwasum,
atm.
Ofa kg./m.'/hr.
GRANULAR CATALYST,
Fe
1 10 10
Fe
190-224 230 275
Fe
250-275
co E'e
175-225 240-260 270-290
Fa
300-320
Fe co
300-320 175-240
Fe
10 20 20
line
Dim1
8 10 10
56 35 32
33 35 18
50 19 68
u)
M)
GRANULAR CATALYST, INTERNALLY 7 40 a5
FC
20 20
11 30 35
22 19
19
40
40
60
60
22 56 8
FLUIDIZED CATALYST, 65 30
v.s.= V.S.0 15
2.7 2.4 2.5
76 61
6 6
6
1.9 2.1 2.2
4
1.2
GAS RECYCLE
32
14
COOLED, QAB EECYCLIO
a5 26 15
30 24 16
GRANULAR CATALYST, HOT-QAB R ~ C Y C W 20 32 70 17 1 20 3
Water solublea 5t.d organiee ton/bbl./dsy
EXTEIRNALLY COOLED, QAS RECYCLE)
13 14 14
POWDERED CATALYST - OIL SLURBY, 20
Wtu
EXTERNALLY COOLED, NO GAB ~ C Y C L B
GRANULAR CATALYST,
co
Gas+
V.8.' 10 10
0.8 0.7 0.6
12
0.7
9
0.6 0.8
QAE RECYCLE
15 38
1 1
V.S.O
a Weight of total product excluding methane, ethane, end ethylene, per volume of catalyst per unit time. b In oonverter and ih acceesories only. C V.8. very small.
-
PROCESSES FOR SYNTHESIS O F HYDROCARBONS
133
eteam in two vessels, each about 2000 cubic feet in volume. The synthesis converters reportedly will be six in number, and each of about 3500 cubic feet volume. They are to be fed with 350,000,000 cubic feet/day fresh synthesis gas and 750,000,000 cubic feet of recycle gas. During the past two years, the Federal Bureau of Mines (65) has developed a process in which the heat of the reaction is removed from a granular or pelleted catalyst bed by the evaporation of an oil introduced along with the synthesis gas. The converter contains no metal heat-exchange surfaces, and the operation is essentially adiabatic. Temperature control is precise, and there is no radial temperature gradient. The vertical gradient is about 25" C. and is a thermodynamic consequence of the decrease in partial pressure of synthesis gas as the reaction proceeds. Most of the operations thus far have been with a cobalt-thoria-kieselguhr catalyst, for which a study was made of the change in space-time yield with space velocity of the 2H1+1C0feed gas. Currently, a similar study is in progress with an iron catalyst. Further operations using recycle of end gas and two or more stages will be investigated to determine the maximum space-time yields. The estimates of Table XI, except those in the first three rows, are based on very meager amounts of data and therefore are only preliminary approximations. As may be seen in Table XI, the internally cooled converter process, in its present (1947) stage of development, has a space-time yield about two-thirds of that of the fluidized iron catalyst process with about 30% more steel necessary for installing of the converter and its auxiliary equipment. The I.C.C. process is quite versatile in that various types of catalyst, and a variety of operating pressures may be used. The product distribution can be varied, therefore, over a wide range.
IV. PROCESSES RELATED TO THE FISCHER-TROPSCH SYNTHESIS The "synol" process for the production of alcohols is identical with the Fischer-Tropsch synthesis using an iron catalyst, differing from it only in space velocity of feed gas and operating temperature (41). Passage of 1CO+0.8Ha+20 to 50 volumes of recycle gas, at 18 to 25 atmospheres and 190-200" C. over a catalyst which was a fused mixture of 97Oj, Fe304,2.5% M2O3and 0.5% KzO previously reduced at about 450" C. with hydrogen, yielded about 160 g./cubic meter of fresh gas of a mixture of hydrocarbons and alcohols. The boiling range and composition of the product are given in Table XII. The catalyst was prepared by burning pure iron in oxygen to form a molten mass of oxide. Aluminum nitrate and potassium nitrate are then added to the melt and the fused mixture cooled and crushed to 1 to 3 mm. particle size. The finished preparation contained 97% Fe304, 2.5% A1203,
134
H. H. STORCH
0.2 t o 0.6% KaO, 0.1670 sulfur, and 0.03% carbon. Its bulk density, after crushing, was about 2.0. ThiR catalyst was reduced for 50 hours in pure dry hydrogen at about 450" C. About 2000 volumes hydrogen/hour are passed through 1 volume of catalyst a t a linear velocity of at least 20 cm. second. The reduced catalyst is transferred in an atmosphere of carbon dioxide to the reactor. T.r\BI,E XI1 ..
Composition of Synol Products
_ _ --
lioilinc r n n m 'C. __. -
34-105 105-150 150-175 175-218 218-255 255-290 29G320 320-360 over 360
Per cent 01
totnl pridiii:t
AlC:ollOl
Aldchydea
~~~
29.7 19.7 5.0 12.7 9.2 6.4 5.5 4.8 2.1
20" 55 44 55 55 49 45 35
4 0 8 4 2 3 3 4
.
.
F a t t y snide
0.2 0.7 0.3 0.3 0.4 0.5 0.5 0.6
2 2.5 4.5 8 10 15 19.5 20
Full-scale commercial development was not completed for the synol process. Twenty-five laboratory to pilot-plant size units were operated, ranging in sizc from 0.2-36 1. of catalyst, to study the relative merits of multistagevs.sing1e-stageoperation and gasrecyclevs. single-pass operation. A four-stage unit of 100 1. catalyst capacity, producing 3-5 tons/month of liquid product was operated for 9 months. The catalyst was contained in 14-mm. i.d. tubes spaced 26 mm. from the center of any tube to the center of a neighboring one. The results of development work in this pilot plant indicated the following operating conditions for a two-stage process: First stage: Feed gas, 20-50 volumes recycle gas/volume fresh gas (1CO:0.8H2); space velocity, 2500 to 5500 volumes total feed gas/volume catalyst/hour ; temperature, 185" C. a t start, 195" C. a t end of 3 months, 225" C. a t end of 9 months; pressure, 20 atmospheres. The gas and vapors from the converter would pass through a hot separator to remove wax, then through a cold separator to remove the alcohol and liquid hydrocarbons, and finally through a recycle gas heater back to the converter. A portion of the recycle gas mould be sent to the second stage, which is identical with the first so far as flow diagram is concerned, except that recycle gas from the first stage is the "fresh-gas" for the second stage and that about one-third of the sccond-stage recycle gas is continuously purged by passage
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
135
from the cold separator through active charcoal and then into the fuel-gas mains. Conversion in the first stage is 65 and in the second 25% of the synthesis gas. To avoid corrosion, alkali must be added just beyond the hot-wax separators. Plugging of reactor tubes due to carbon deposition was the chief operating difficulty. The synol process is of interest in a discussion of the Fischer-Tropsch reaction for the following reasons: (1) The same catalyst, when used at a higher temperature as in the hot-gas recycle process, yields chiefly hydrocarbons, with only 5-1070 alcohols, whereas in the synol process a very high recycle ratio of about 100 with drying of the gas on each cycle yields as much as 70% conversion to alcohols. The total conversion is over 90% to alcohols plus hydrocarbons, with only a very small amount of carbon dioxide produced. I t is likely that the formation of the normal straight-chain alcohols, which constitute the bulk of the synol alcohol product, precedes the formation of olefins on this iron catalyst at 20 atmospheres in the temperature range 190-325" C. (2) Carbon deposition apparently is proceeding at an appreciable velocity even at 190-225" C. under the conditions of the synol process. The formation of carbon at 300-325" C. in the hot-gas recycle or the fluidized iron catalyst processes, therefore, is not peculiar to the elevated temperatures used in those processes, although its rate is higher at the higher temperatures. The "0x0" process consists in reacting olefins with carbon monoxide and hydrogen to produce aldehydes, and subsequent reduction of the aldehydes with hydrogen to alcohols. The basic reaction was discovered by the Ruhrchemie, and a large-scale continuous process resulted from co-operative effort of Ruhrchemie and I. G. Farbenindustrie (41, 66). A slurry of liquid olefin plus 3-5y0 reduced Fischer-Tropsch cobalt-thoriamagnesia-kieselguhr catalyst is pumped along with l H z + l CO gas (about 0.2 cubic meter/ 1. olefins) through a preheater (150-160°C.) into the reactors at 200-atmospheres pressure. The first reactor is 20-cm. i.d. by 8-meters high, it contains six 21-mm. 0.d. by 17-mm. i.d. cooling tubes connected to a cooling jacket around the converter. The second reactor is in series with the first and contains some baffles but no cooling tubes, and an equal amount of gas (0.2 cubic meter/l. olefin) is pumped into the second reactor. The excess H,+CO in both reactors serves to agitate the reaction mixture. Ethylene and propylene react readily when used in solution in a liquid medium. The contact time is about 20 minutes. Most of the catalyst is filtered off and recycled. The conversion is 70% in the first reactor and 30% in the second. The product consists of 80% aldehydes (about onequarter of this is polymerized aldehyde) and 20% alcohols. It is not practical to isolate aldehydes from the first stage and it is considered prefer-
136
H. H. STORCH
able to dehydrogenate the corresponding alcohols from the second stage (see below). Because the hydrogcnation of the aldehydes is retarded by carbon monoxide the first-stage product must be let down to atmospheric pressure. The second stage is operated with hydrogen at 170-195” C. and at 200atmospheres pressure. Fischer-Tropsch cobalt catalyst or copper chromite may be used in the second stage. If the former is used, some carbon monoxide is formed by hydrogenation of cobalt carbonyl, and the exit gas is passed over an iron catalyst at 250” C. to convert the carbon monoxide to methane whose concentration, up to about lO%,is of only slight importance. The second stage is otherwise identical with the first. The first stage is not affected by sulfur compounds, but the second is quite sensitive; and if sulfur compounds are present, sulfide catalysts (such as a mixture of nickel and tungsten sulfides) must be used. The reaction in the first stage probably is a homogeneous catalytic reaction, with either cobalt carbonyl or cobalt hydrocarbonyl as thecatalyst. The insensitivity to sulfur poisoning, and the fact that any finely divided reactive cobalt compound may be used, agrees well with this assumption. The aldehydes and alcohols produced are a mixture of normal and isocompounds. This is due not only to the orientation of the hydrogen with respect to the C-CO bonds in the initial reaction complex but also to the isomerization of the olefin under the process conditions. It may be significant that nickel carbonyl does not readily shift the olefin double bond under the “0x0”process conditions, and nickel compounds are very poor catalysts for the process. From isooctene 32% n-nonyl alcohol, and from propylene 507, n-butyl alcohol are obtained, the remainder of the products being isoalcohols. In general, using a-olefins a3. raw material, one obtains about 60% isoalcohols. The synthesis will not occur unless a labile hydrogen atom is available in the olefin reactant. With diolefins the reaction takes place at only one double bond. It is possible that a reaction similar to the “0x0”reaction occurs on the surface of catalysts during the synthesis of hydrocarbons and alcohols by the Fischer-Tropsch and the “synol” processes, and accounts for the small (about 10%) fraction of branched hydrocarbons and isoalcohols produced. In this connection it is of interest to note that as Pichler (37) indicated, the optimum operating pressures for nickel, cobalt, iron, and ruthenium Fischer-Tropsch catalysts increase in the order mentioned, and the d s c u l t y of metal carbonyl formation also increases in this order. The isoparaffin synthesis studied at the Kaiser Wilhelm Institut fur Kohlenforschung (37, 67,68, 69)is more closely related to the synthesis of the higher molccular weight alcohols using difficultly reducible oxides plus alkali as catalysts, than to the Fischer-Tropsch process. Thoria, Binc oxide,
PROCESSES FOR SYNTRESIS OF HYDROCARBON0
137
alumina, or a mixture of these is the catalyst for the isoparah synthesis. The operating temperature is about 450' C. and pressure 300 atmospheres or higher of 1C0+0.85Hz. Figures 3 and 4 show the effect of operating
FIG.3. Reaction product composition as related to temperature. (Tho, single component catalysts)
0
300 600 PRESSURE, ATMOSPHERES
900
Fro. 4. Influence of pressure on the amount and kind of products from the isosynthesis. (Thozsingle component catalysts, 450" C.)
138
H. H. STORCH
temperature and pressure on the composition of the product using thoria as catalyst. At 300 atmospheres and 400450" C., using thoria plus 20y0 alumina and 3% potassium carbonate as catalyst and about 1000volumes (measured at atmospheric pressure) of synthesis gas per volume of catalyst, 70% conversion per pass was obtained. The yield in g./cubic meter of synthesis gas was : Liquid hydrocarbons 42 1s0-C~hydrocarbons 51 n-C4Hydrocarbons 7 Ca Hydrocarbons 18 COHydrocarbons 6 CHI Hydrocarbons 12 Water soluble alcohols 4 Oil soluble alcohols 2 The olefin and aromatic content of fract,ions of the liquid hydrocarbons increased with increasing boiling range of the fractions. Thus the 65-95" C. fraction contained 25% olefins, and no aromatics, and the 175-195" C. fraction contained 62% olefins and 31y0aromatics. After saturation of olefins and aromatics by catalytic high-pressure hydrogenation at low temperature, the liquid product contained 34% C6, 21y0 Cs,10% C,, 5oJ, Cs paraffins, naphthenes, and 11% higher-boiling hydrocarbons. The Cg fraction consisted of 91% 2-methylbutane and 9% n-pentane. The Cs fraction contained 3% 2,2dimethylbutane, 47% 2,3-dimethylbutane1 31% 2-methylpentane, 16% 3-methylpentane, and 3% methylcyclopentane; n-hexane was not detected.
V. CARBIDEFORMATION AND CARBONDEPOSITION ON IRON CATALYSTS The formation of carbon on iron and iron-copper catalysts by the reaction 2CO=COz+C has been studied by several investigators (70-73). The most significant result of this work (in so far as the Fischer-Tropsch synthesis is concerned) is the fact that neither an iron-free nor a copper-free carbon deposit was obtained. The data show that carbon is deposited in the crystal lattice of the catalyst and the inability to obtain a copper-frcc carbon deposit from tests with an iron-copper catalyst shows that iron carbonyl formation will not explain the results. It is very probable that carbon is formed from carbon inonoxide by wny of iron carbide as an intermediate. Carbidic carbon diffuses rapidly throughout the crystal lattice and subsequently decomposes to yield elemental carbon, thus disrupting the lattice structure. Pichler and Merkel (74) at the Kaiser Wilhelm Institut fiir Kohlen-
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
139
forschung showed that extensive conversion of iron catalysts to the carbide Fe2C results in a highly active and durable Fischer-Tropsch catalyst. Thus, treatment with carbon monoxide at 1/10 atmosphere and 325" C. of precipitated iron catalysts yields a maximum of carbidic carbon of 4.670, which corresponds to maximum activity and durability in the synthesis at 235" C. and 15 atmospheres pressure. The carbidic carbon content decreased to about 3.5% during the first few days of the synthesis but remained constant for months thereafter. The oxygen content of the catalyst was 0.6% at the start and increased to 1.6% a t 112 days. In catalysts operated for long periods up to 550 days at 210 (start) to 285" C. (end temperature) the carbidic carbon dropped to 1%,the oxygen content increased to more than 20% and the elemental carbon to 8%. High H2:CO ratios in the feed gas reduce the rate of carbon formation. Alkalized iron catalysts form elemental carbon more rapidly than those containing little or no alkali. Iron carbide (FeZC) of Curie point 265" C. is formed almost exclusively by treatment of precipitated, unreduced iron cat,alysts with 0.1 atmosphere of CO at 325" C. This carbide is only very slowly hydrogenated in pure hydrogen at 230-240' C. Promotion of precipitated catalysts speeds FezC formation. There is some evidence for the existence of a carbide of Curie point 380" C., which is obtained by treating a reduced (at 250" C.) iron catalyst with carbon monoxide at 220" C. A catalyst pretreated in this fashion is of low activity in the Fischer-Tropsch synthesis. Both carbides of Curie points 265 and 380" C. are converted to cementite (Fe3C) upon heating at temperatures higher than 300" C. Precipitated iron catalysts which are not pretreated, but used directly in the synthesis at 235" C. and 15 atmospheres, show a slow increase in Fe& content. Thus after 120 hours the ratio Fe&:Fe304 is 3:lO. The activity in the synthesis increases to a maximum during this period. If one assumes that elemental carbon formation occurs by way of some form of iron carbide as an intermediate, the fact that increasing the Hz:CO ratio in the synthesis gas reduces the rate of carbon formation may be explained in either of two ways: (1) The hydrocarbon synthesis and the elemental carbon formation are competing for the available supply of carbide. Higher partial pressures of hydrogen are known to increase the rate of hydrocarbon synthesis and therefore the rate of elemental carbon formation would be expected to decrease with increase of Hz:CO ratio. This explanation assumes that bulk carbide (as distinguished from surface carbide) is a necessary intermediate in the Fischer-Tropsch synthesis. As shown by the recent work of Emmett and his associates (75) using radioactive carbon as a tracer, this assumption is very probably incorrect.
H. H. BTORCH
140
(2) The formation of carbon to carbon bonds in the lattice of the catalyst must occur to obtain elemental carbon from metal carbide. If the diffusion of carbidic carbon to nuclei where such carbon to carbon bonding has started is retarded by the penetration of the lattice by atomic hydrogen , the rate of elemental carbon formation would decrease with increasing partial pressure of hydrogen. This explanation, based on rate of diffusion of carbidic carbon in the lattice, may serve also to account for the effect of alkali and other promoters or impurities. Thus the presence of alkali tends to preserve a structure similar to that of the spinels, and it is possible that carbidic carbon diffuses more readily through such a lattice than through that obtained when little or no alkali is present.
VI. KINETICB AND REACTION MECHANISMB IN THE FIBCHER-TROPBCH SYNTHEBIB 1. Pressure and Temperature Coeficients. Effect of Contact Time
The early work of Fischer and coworkers showed that pressures above 1atmosphere favor the formation of oxygenated organic compounds and of high-molecular-weight hydrocarbons, and reduce the yield of normally liquid hydrocarbons per cubic meter of synthesis gas (76). Pressures lower than atmospheric necessitated proportionately greater amounts of catalysts for equal conversion to hydrocarbons, although the average molecular weight and chemical composition of the product were unaffected by the reduced pressure (77). In the absence of diluents, the reaction rate is approximately directly proportional to the pressure of 2HaS-1 CO in the pressure range from 1 to 0.1 atmosphere. TABLE XI11 Effect of Nitrogen Dilution rn Yield (Cobalt catslyat, atmospheric pramre, 100 volumes of 2Hz+l CO/volume of catalyet/hour) Nitrogen, %
Yield, E. oil/m.' CdcuQted
Observed
1.0
101
16.3
88 90
28.8
... 84.5
72.3
____.
On a cobalt catalyst a t a total pressure of about 1 atmosphere, dilution with nitrogen does not decrease the percentage of conversion. In fact, such dilution results in a percentage of conversion somewhat greater than would
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
141
be calculated upon the basis of a first-order reaction. Storch and his coworkers (55) report the data in Table XI11 on nitrogen dilution. Similar results have been reported by other investigators (78,79). The facts that in the absence of diluents the reaction is approximately first order andthat in the presence of nitrogen the reaction rate is somewhat greater than corresponds to the partial pressure of the synthesis gas make it probable that the slow step in the reaction is the desorption of the products. The retarding effect of the reaction products is illustrated by the data of Aicher, Myddleton, and Walker (80) who used a nickel catalyst: “If adsorption (of the products) is an important retarding influence on the reaction, the rate of reaction should vary directly as the partial pressure of reactants I
I
0
9 1
B
PF
: Z I
B
3
3 i!
I
FIG.5. Space-time yield aa function of partial pressure, A , of reactants and partial pressure, I,of products;at 1 atmosphere total pressure.
and inversely as the partial pressure of the product. The yield of oil per unit time, per unit volume catalyst, would then bear a linear relationship to A / x , where A is the concentration of the reacting gas and x is the con-
142-
H. H. BTORCH
centration of the product. By constructing graphs relating length of catalyst to yield, a t various temperatures, and integrating these curves, we were able to estimate the average concentration of unchanged reactant in the experiments, over any catalyst layer." Figure 5 is a reproduction of the data referred to in the quotation from Aicher et al. (80). The straight line for 205" C. fits the data reasonably well, and one may conclude that the rate is retarded by the adsorbed products of the reaction. The data for 190" C. (Fig. 5) are inadequate to determine a straight line, and those for 186' C. indicate that at low conversions there is some deviation from the rule of inverse proportionality to the partial pressure of the reactants. Some of the data of Aicher and coworkers (80) (Fig. 5) are suitable for calculating the temperature coefficient for relatively short contact times (that is, conversions of about 1050%). The coefficient for the temperature ranges 197-207' C. and 191207" C. are 1.4 and 1.67/10° C., respectively. These coefficients correspond to an activation energy pcr mole of about 20 kcal. as calculated from the Arrhenius equation. This energy value is reasonable for a desorption process. Weller (81) studied the pressure dependence of thc synthesis over a cobalt-thoria-kieselguhr catalyst, at about 1 atmosphere pressure using very short contact times so that the concentration of synthesis products was very low. Under these conditions the synthesis rate increased approximately as the square root of the synthesis gas pressure. The temperature coefficient in Weller's tests corresponded to an activation energy of 26 1tcal.l mole. Storcli and his coworkers (55) obtained a value of 25 kcal./mole at long contact times corresponding to the Huhrchemie process. TABLE XIV Efect of Pressures A bow A b i n s p h ~ r i c -
~~~ ~
l'otnl solid
Pressure, ntm.
ond liquid hydrocnrlrons
~
1.0 2.5 6.0 16.0 51.0 151.0
117 131 150 145 138 104
10 15 60
70 54 27
~
~~~~~
Tii*lds, u./culric nieter synthesis gna Oil Gosolinc boiling boiling Parufin nlrovrr lidow WRX 2000 c. 200' c.
38 43 51 38 37 34
69 73 3!) 30 47 43
Gaseous hytirucnrbons _-
CI to c4
38 50 33 33 21 31
A study of thc cfkct of prcssurcs above atmospheric was made by Fischer and Pichler (82). As the pressure was increased above ntmos-
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
143
pheric, the yield at first increased and then (at about 15 atmospheres) decreased. Table XIV shows the results of their work. A cobalt-thoriakieselguhr catalyst and a constant throughput of 1 1. (measured a t atmospheric pressure)/hour of 2H2+lCO gas/ g. cobalt metal were used. The data are averages for 4-meek operation. Only a single pass of the gas through the catalyst was made, and there was no catalyst regeneration. Because of tge constancy of throughput, the contact time increased in direct proportion to the pressure. The yield of all hydrocarbons (column 2 plus column 6) increased from 155 g./cubic meter at atmospheric pressure to 181 at 2.5 and 183 at 6 atmospheres, and then decreased with further pressure increase. As the theoretical yield was about 208 g./cubic meter, the maximum percentage of conversion to hydrocarbons was 88-89. The rapid decrease in yield of hydrocarbons for presloo sures above about 16 atmospheres was probably due to corrosion of CN active catalyst surface by carbonyl + formation and by the stronger adsorption of t h e oxygenated " organic compounds which were produced in larger amounts at the higher pressures. Martin's curves (32), illustrating the effect of pressure, are f lo reproducedinFig. 6forcomparison 0.1 02s 0.s 1 0 25 so 10 PRESSURE IN Am. A W T L with data on the effect of pressure on a ruthenium catalyst by pjchler FIG. 6. Effect O f pressure OD total yield and yield of various fractions.
and Buffleb (83) at pressuresfrom 15 to 1000 atmospheres. The results of the latter work are shown in Fig. 7. Three grams of ruthenium catalyst was used, and the flow of synthesis gas was so adjusted that 11. (measured at l-atmosphere pressure) of effluent gas was obtained per hour. The throughputs of 2HZ+l CO (calculated from the nitrogen contents of the ingoing and effluent gases) were 1.08, 1.16, 1.79, 2.57, 5.0, and 5.26 I./hour/3 g. ruthenium for 15, 30,50,100,180, and 1000 atmospheres, respectively. I t must be remembered that the data of Fig. 6 were obtained a t a constant throughput per gram of catalyst, whereas those of Fig. 7 were measured at varying throughput. If the throughput had been kept constant at, for example, 1 l./hour/g. ruthenium, then the yield and conversions would have been appreciably lower at 15, 30, and 50 atmospheres, only slightly lower at 100 atmospheres, and slightly higher at 180 and 1000 atmospheres. Therefore, the curves of Fig. 70n a constant throughput basis would have been steeper between 15 and 180 atmospheres. With
144
H. H. STORCH
this in mind, a comparison of Figs. 6 and 7 shows that, for a cobalt catalyst, about 95% conversion is reached at 15 atmospheres, whereas for a ruthenium catalyst 95% conversion necessitates a pressure of about 300 atmospheres.
lLogarillirnic scale1
FIG.7. Yicld and character of products at 180' C. and various pressura.
It is interesting to note also that,whereas for a cobalt catalystmaximum life was obtained a t 5-15 atmospheres pressure, only a few weeks of operation a t 50-150 atmospheres sufficed to cause a large decrease in catalyst activity; for a ruthenium catalyst 6 months of operation at 100-1000 atmospheres did not result in any observable decrease in activity. The effect of increasing space velocity at pressures above atmospheric is demonstrated by the data of Table XV calculated from the results of Fischer and Pichler's work (82). The space-time yield (column 5) reaches a maximum at about 4 1./ hour/g. cobalt. In column 6 the space-time yield was calculated aa indicated in the footnote. The volume change upon reaction to form chiefly hydrocarbons with 5 to 16 carbon atoms is approximately constant over a wide range of variation of molecular weight distribution; therefore the partial pressure of the reaction products is approximately directly proportional to the contraction. The figures in column 6 increase more slowly than those of column 5 with increasing throughput, but there is no maximum. It seems probable that the temperature of the catalyst surface increases
PROCESSES
FOR SYNTHESIS OF HYDROCARBONS
145
with increasing throughput, despite the apparent constancy of the average temperature. At the higher temperatures a large proportion of the carbon monoxide used appears as gaseous hydrocarbons and as carbon dioxide. TABLE XV Efect of Varying the Throughput at 15-Atmospheres Pressure and SOOO C., wing .G g. Cobalt (in Cobalt-Thoria-Kieselguhr) 2
3
L./hr./g. Co
Contraction, %
1 2 Ha
+ 1CO
L./hr.at 1 atm.
4.0
1.15
74
9.0
2.25
60 48 16
3.6
14.6 40.5
10.0
4
6
101 85
192
__
6
Yield of oil (ercludinn nsaol). G./1000 G./1000hr g. hr./g. Co co. caloa: O./rn.' (col. 4 x 2 ) from oontractiona
60 16.5
106
106 131
216
164
165
490
Anaud that the volume ohange upon reactipn M constant for dl througbputa and that yield is independent2 the partial pressure of reactanta and inversely proportional ta partial pressun, of products. Thus taking 106 g./lOOo houm 89 the bssis for comparison the second-row figure of oolumn 6 is obtained from'lO6 X 0.74/0.6, the third row from 108 X 0.74/0.48, A d the fourth from 108 X 0.74/0.16
Additional data on the effect of varying space velocity in the range of 0.24 to 2.16 1. 2Ha+l CO/g. cobalt, operating at 10-atmospheres pressure and using a cobalt-thoria-kieselguhr catalyst, are given by Fischer and 2w
150
n
ELL
1w
VI
I
a 50
C -
I I 1 lpal 1
: 0.4"
I
" 16 20 THROUGHPUT, LITERS PER HOUR PER G R A M Co
FXQ.8. Intermediate pressure synthesis yields and carbon balance MI.throughput.
Pichler (84). In each of these experiments the operating temperature was adjusted to obtain a minimum of methane and maximum of liquid plus solid hydrocarbons. The results are presented in Fig. 8, where the out-
146
H. H. GTORCIH
standing fact is the extraordinarily small amounts of C2 hydrocarbon produced. This marked discontinuity in the quantitative composition of the product is of interest in connection with the data given in an earlier section of this paper concerning the unique position of CzH4in t h e synthesis on cobalt catalysts. Figure 8 shows that the yield of solid paraffi decreased with increasing space velocity, as did also the total yield of hydrocarbons other than methane. Hecause the data of Fig. 8 were obtained by separation of the condensable (at room temperatures and 10-atmospheres pressure) hydrocarbons in two stages, and because only the total amount of condensed hydrocarbons for l o t h stages is given, it is not possible to correlate partial pressure of the products with space-time yield. The effect of varying space velocity over a ruthenium catalyst a t 230" C. and 100-atmospheres pressure as given by Pichler and Buffleb (85) is presentcd in Table XVI. The space-time yield shows a maximum TABLE XVI
__
Effect oj Space Velocity on a Ruthenium Catalyst ut 230°C.and 100-utmospheres pressure
__
1
________-
______I_-________
2
3
6
4
La./hr.; Contrar- Cubic iiictcr G. Ru/hr. XlW g. Ru tion, % ' 2 H,tl CO 1.8 3.2 4.3
7.0 9.1 10.7 16.7
81 79 69 62 96 50 36
136 125 98 84 79 65 37
24 40 43 59 72 69 60
7
6
CO Converted
Yield, g. liquid per
L. catalyst apereflir.
8 TotaI hydro-
to condensed carbons, hydrocnrhons _ _ g.11. cntalyet Per cent Per annt spacelhr. CO used CO reacted
108 180 194 265 324 310 270
87 83 74 70 73 68 54
78 72 56 48 45 37 21
-~
______ .~~ ~
124 217 262 379 445 456 500 ~- ..
a t a throughput of about 9 1. 2Ha+lCO/hourig. of ruthenium. The space-time yield (column 5) is plotted agttinst the percentage of contraction (column 2) in Fig. 9. It was noted in column 5 of Table XV for a cobalt catalyst a t 15 atmospheres that the space-time yield goes through a maximum a t about soy0 contraction. In Fig. 9 a similar maximum a t about 55% contraction is observed for ruthenium at 100-atmospheres pressure. Apparently the heat transfer capacity of the apparatus used was exceeded as the throughput was increased. The resulting rise in temperature of part of the catalyst bed led to an increased Fpace-time yield of gaseous hydrocarbons and a correspondingly lower yield of liquid hydrocarbons. Column
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
147
7 of Table XVI shows that, with increasing space velocity, the fraction of the reacting carbon monoxide which yielded condensed hydrocarbons decreased. As only negligible quantities of carbon dioxide are obtained (85) virtually all of the carbon monoxide which reacted yielded hydrocarbons. From columns 5 and 7 of Table XVI may be calculated the approximate
FIG.9. Synthwis on ruthenium. Relation of spacetime yield to contraction.
space-time yield of all hydrocarbons (solid, liquid, and gaseous). The results of such calculations (column 8) show no maximum yield such as is observed in column 5 and Fig. 9 for the yield of condensed hydrocarbons. Pichler and Buffleb (85) also published some data on the contractions and yield obtained at several temperatures from a ruthenium catalyst at a constant space velocity of 7 1. 2H2+ 1CO/g. ruthenium and at the higher pressure of 1000 atmospheres. These data are given in Table XVII. The space-time yield of condensed hydrocarbons is plotted against contraction in Fig. 9. Within the limits of heat-transfer capacity of the apparatus, the yield is directly proportional to the contraction when the temperature is varied, because the temperature coefficient of the reaction rate apparently is large enough to overcome the increased retarding effect of the products with increasing conversion.
148
H. H. BTORCH
The effect of pressures above 1 atmosphere on the durability of a cobalt catalyst was investigated by Fischer and Pichler (82). Their results are presented graphically in Fig. 10. The pressures given are gage readings, so that the O-atmosphere curve shows the results at l-atmosphere pressure. TABLE XVII Egect of Temperature m Yield for Ruthenium Catalyst at 1000 Atnwspheres
____
Temo. 'c.
Contraction. %
164 180 190 200 210
49 55 62 65 69
Yield of liquid CO Converted md solid hydrooarbons, to c o n d e d hydrocarhorn, g.irn.8 per cent CO used
72 79 89 95 104
41 44 51 55
80
The yield of solid plus liquid hydrocarbons showed a minimum decrease with time for reaction pressures of 6-16 atmospheres. At these pressures the initial yield waa more than 150 g./cubic meter, and after 26 weeks of operation the yield was still above 100 g./cubic meter. At 1-atmosphere pressure the yield dropped from 130 to 100 g./cubic meter in only 4 weeks of operation, At 151 atmospheres the decrease in yield with time waa very rapid. At such pressures there wm a noticeable formation of cobalt carbonyl and consequent corrosion of the catalyst surface.
OPFXATINO PERIOD (WEEKS1
Fro 10. Variation of yield with time at various prwures.
The slower decrease in activity of the catalyst at medium pressures wag shown also by the smaller temperature increase necessary to obtain the maximum yield of liquid and solid hydrocarbons. Increasing time of operation was associated with gradual lowering of the activity of the catalyst. This reduction could be partly compensated by raising the reaction temperature. In the experiments at 6-16 atmospheres the initial temperatures of 175-180" C. were not raised to 195" C. even after 6 months of operation. At 1 atmosphere the initial temperature of 180" had to be increased to 196" C.after only 5 weeks of operation.
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
149
Similar data on the effect of pressure on the durability of iron catalysts are presented by Pichler (37,74,86). Iron catalysts usually are much less durable at atmospheric pressure than are cobalt catalysts. As with the cobalt catalyst it is possible, however, by frequent flushings with hydrogen to keep a carefully prepared and pretreated iron catalyst active at atmospheric pressure for 6-12 months (37). The pressure coefficient of the synthesis on iron catalysts is shown in Fig. 11, prepared by Pichler (37)from 175 K 3 I 0 0: P W
150
-
-
v) I-
53 125
-
0
W K
c2
I00
W K
a
0 c W
75
8 v)
u a
50
-
8 ul W
5 250 PRESSURE, ATMOSRIERES
FIG.11. G m conversion on an active iron catalyst a t various pressures. Extensive carbonyl formation above 20 atmospheres.
data obtained a t the Kaiser Wilhelm Institut fur Kohlenforschung. This pressure coefficient decreases with decreasing operating temperature and a t about 175" C. probably would be close to that of a zero order reaction as found for cobalt catalysts. 2. Mechanism of the Synthesis F. Fischer's hypothesis concerning the mechanism of the synthesis postulated the intermediary formation of cobalt or iron carbide and reduction of the carbide to methylene (CHS groups. The latter would subsequently polymerize to yield unsaturated paraffin hydrocarbons. Although direct hydrogenation of preformed cobalt or iron carbide yields largely
150
H.R. BTORCH
methane, with only minor :mounts of other hydrocsrbons, the hypothesis of Fischer could be correct because of the selective poisoning action of carbon monoxide and of the products of the synthesis in retarding the methane forming reaction. One of the earliest significant observations relating to the mechanism of the reaction was made in 1930 by Dr. D. F. Smith and his coworkers (16). They found that when ethylene was added to synthesis gas (2H2+1CO) and the mixture passed ovcr a Co-MnO-Cu catalyst a t about 200” C., the ethylenc entered into the synthesis thereby increnaing the yield of oil and of particularly of oxygenated organic compounds. When, however, an Fe-Cu catalyst mas used, the oil yicld was virtually unaffected by the presence of ethylene in the synthesis gas. This obswvation \vas directly confirmed by Craxford (17) in 1939. Dr. C. C. Hall of the British Fuel Research Laboratory (87) finds that when adsorbed wax on a cobalt catalyst is removed by trcatment with hydrogen at about 210’ C. the chain-length distribution curvc for the products has the same peculiar form as the corresponding curve for the synthesis products. Thus, there is a minimum at Ca for both synthesis and wax hydrocracking, n maximum at Cg to Cs for the synthesis products, andat CstoCgfor thc cracking products followed by an asymptotic approuch to zero at higher molecular weights. This suggests to IInll the hypothesis that during synthesis, the array of interacting CI& groups on the catalyst surface may be regarded as a giant molecule and that observed synthesis products are fragments split off this by the attack of chcmisorbed hydrogen atoms. When the surface concentration of the latter is low, little cracking will occur, and long chains will be produced; and as it increases more and more cracking mill occur, the products will become shorter and shorter in chain-length, and finally only methane will be produced. Data on secondary reactions such as hydrocarbon cracking and the water-gas shift occurring on cobalt catalysts have been published by Craxford (88). The conccpt of metal carbide aa a synthesis intermediate has been used rather loosely. Distinction has rarely been made between bulk carbide and “siirface” citrbide as a possible intermediate. Craxford and Rideal (89), fur example, speak of ‘(si~rfacc’~ carbide; they give, however, aa evidence for the carbide theory only examples which indicate that under some conditions bulk carbide accumulates in the catalyst. “Surface” carbide is very difficult to distinguish from an adsorbed reaction complex of hydrogen and carbon monoxide. Weller, Hofer, and Anderson (90) present evidence that bulk carbide, in the case of cobalt catalysts, probably is neither an intermediate in the synthesis nor a catalytically active substrate for the synthesis. They show that presence of extensive amounts of carbide in cobalt catalysts severely inhibits the Fischer-Tropsch synthesis. In one experiment, for example, a cobalt-thoria-magnesia-kieselguhr (100:6:12:2Oo)
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
151
catalyst was reduced with hydrogen at 400" C. and carbided with carbon monoxide for 20 hours at 200" C.; the amount of carbon monoxide reacted was equivalent to 70% conversion of the cobalt to Co&. Exposure of this carbided sample to circulating synthesis gas resulted in an overall synthesis g. carbon/g. cobalt/hour. After removal rate at 150" C. of 60-70 x of the carbide by hydrogenation at 150-194°C. and re-exposure of the sample to synthesis gas, a synthesis rate of 250-300 x g. carbon/g. cobalt/hour was observed, an increase of 4000j0. It may be noted that the distribution of synthesis products was about the same in both cases. In another case, a sample of cobalt-thoria-kieselguhr (100:18:100) catalyst which had been completely carbided a t 210°C. was found to exhibit no appreciable activity in the synthesis under conditions where a control sample, similarly reduced initially but not carbided, gave copious yields of oil and water. Partial hydrogenation of the carbided sample at 190" C., with the removal of 72% of the carbon which had been introduced, restored the activity to about one-half the normal value. It was further shown that the loss of synthesis activity on carbiding can persist even when synthesis is continued for several days. Another sample of cobalt-thoria-kieselguhr catalyst was carbided for 24 hours at 200" C. after initial reduction. On exposure to flowing synthesis gas at 180" C., the carbided sample originally showed no catalytic activity, as measured by the fractional decrease in gas volume (contraction) as the gas passed through the converter. During operation for 1 day at 180" C., the contraction rose to 20%, but it did not increase further during another day of synthesis. Increasing the temperature to 190" C. increased the contraction only to 32% even after a period of a day. Subsequent hydrogenation of the sample at 210' C. and re-exposure to synthesis gas resulted in a normal contraction of 60% at 180°C. Recarbiding of the active catalyst for 24 hours at 200" C. again reduced the activity to the low value observed after the first carbiding, and rehydrogenation of the carbided sample again restored the normal activity. It should be noted that this inhibitory effect of precarbiding before synthesis is not explainable as a poisoning due to free carbon deposited during the carbiding, because the effect is completely reversed by hydrogenation at temperatures where free carbon is not affected. These experiments show that bulk cobalt carbide is not a catalyst for the synthesis; this conclusion, furthcrmore, is contrary to that which would be expected if bulk carbide were an intermediate in the synthesis. Further evidence which indicates that bulk cobalt carbide is not of importance in the synthesis is furnished by x-ray analysis of catalysts which have been used in the synthesis. It has been shown (91) that on low-temperature carburization with carbon monoxide, reduced cobalt
152
H. H. STORCH
catalysts form a carbide, Co2C,which has a characteristic x-ray diffraction pattern. This fact provides a means of identifying cobalt carbide in the catalyst. It was then observed by x-ray analysis that if catalyst mmples were partly carbided before synthesis and then used in synthesis, carbide was still present in the samples even after several days of synthesis. (In pure hydrogen, the carbide would have been completely removed as methane in several hours.) On the other hand, if the samples were used in the synthcsk immediately aftcr reduction (that is, no precarbiding), no carbide was ever found at the conclusion of the synthesis. In other words, bulk carbide is apparently not built up in appreciable quantity during steady-state synthesis, while carbide already present is apparently not removed a t an appreciable rate during steady-state synthesis. Although quantitative determinations of the amount of carbide present under given conditions have yet to be made, the indication is that bulk carbide is neither formed nor destroyed as part of the synthesis process. The final evidence to be adduced in connection with the role of bulk carbide is also based on x-ray analysis. It was shown (90) that cobalt in catalysts reduced at 360400" C. is present primarily as cubic cobalt. Cubic cobnlt is thermodynamically unstable with respect to hexagonal cobalt at temperatures below about 360" C. Curbiding of the reduced sample and hydrogemtion of the carbide, both at 200" C., always converts the cobalt to the low-temperature stable, hexagonal form. Now all samples of cobalt catalysts (not precarbided) which have been examined aftcr use in the synthesis have given a diffraction pattern identical with that of the freshly reduced catalyst. If any appreciable amount of the cobalt had been through the cycle of carbiding and hydrogenation at any time during the synthesis, it would have appeared as hexagonal, not cubic, cobalt in the uscd catalyst sample. Since this did not occur, it follows that, within the limits of the x-ray method, the cobalt was not converted to bulk carbide and subsequently hydrogenated during the synthesis; that is, bulk carbide is not an intermediate in the synthesis. Experiments have recently been completcd by Kummer, DeWitt, and Emmett (75), using C'* as a tracer in the synthesis on an iron catalyst. The results are inconclusive. If the total catalyst surface is uniformly active in the synthesis, the results show that only a small fraction of the reaction procecds by way of the carbidc. However, if only occasional active patches of the surface are participating in the synthesis, then it is possible to interpret the results as indicating that all of the reaction proceeds by way of the carbide. The carbide intermediate hypothesis for the mechanism of the synthesis on iron catalysts, however, is probably incorrect. Thus, the results of recycle operations at low tempcratures on iron catalysts show that alcohols are formed earlier in the synthesis than olefmic hydrocarbons.
PROCESSES FOR SYNTHESIS OF HYDROCARBONS
153
This fact is particularly difficult to explain by the carbide-methylene polymerization hypothesis. It appears probable that the olefins are produced by dehydration of the alcohols and that the latter are very early products of the synthesis. The fact that only a relatively small fraction of singly-branched hydrocarbons is produced in the synthesis on iron and cobalt catalysts is very difficult to esplain on the basis of the carbide-methyIene polymerization hypothesis. Branching should occur whenever a carbidic carbon is included in a methylene polymerization chain, and as there is according t o this hypothesis a considerable supply of carbidic carbon, branching should occur much more often than is the case. Furthermore, there is no explanation offered by this hypothesis as to why the branch is limited to a single methyl group. If alcohols are the precursors of the hydrocarbons in the synthesis, it is probable that a reaction analogous to the “0x0”synthesis occurs on the surface of Fischer-Tropsch catalysts. Such n reaction would account for the relatively small fraction of branched hydrocarbons and for the fact that the branches are almost entirely single methyl groups.
REFERENCES 1. Badische Anilin u. Soda Fabrik, German Patent No. 293787 (1913). 2. Fischer, F., and Tropsch, H., Brennstofl-Chem. 4, 276-85 (1923). 3. Fischer, F., and Tropsch, H., German Patent No. 484337 (1025); Btenmtoff-Chem.I , 97-104 (1926); Ber. 69, 830-1, 923-5 (1926). 4. Fischer, F., Proc. Intern. Conf. Bituminous Coal. 1 s t Conf. 231-46 (1926). 5. Elvins, 0. C., and Nash, A. W., Fuel 6,263-5 (1926); Nature 118, 154 (1926). 6. Nash, A. W., J . SOC.Chem. Z n d . London 46,876-81 (1926). 7. Smith, D. F., Davis, J. D., and Reynolds, D. A., Ind. Eng. Chem. 20,462-4 (1928). 8. Fischer, F., and Tropsch, H., Brennsbff-Chem. 9 , 2 1 4 (1928). 9. Tropsch, H., Ges. Abhandl. Kenntnis KohZe 9, 561-70 (1928-9). 10. Tropsch, H.. and Koch, H., Brenmtofl-Chem. 10,33746 (1929). 11. Smith, D. F., Hawk, C. O., and ReynoIds, D. A., I d . Eng. Chem. 20,1341-8 (1928). 12. Kodama, S., Sci. Papers Inst. Ph?p. Chem. Research (Tokyo) 14, 169-83 (1930). 13. Kodama, S., and Fujimura, K., J . SOC.Chem. Id.,Japan 34, Suppl. Binding, 14-6 (1931). 14. Fujimura, I<., J . SOC. Chem. Znd., Japan 34,136-8, 227-9,384-6 (1931); 36, 179-82 (1932). 15. Fujimura, K., and Tsuneoka, S., J . SOC.Chem. Znd., Japan 36, Suppl. Binding, 415-6 (1932). 16. Smith, D. F., Hawk, C. O., and Golden, P., J . Am. Chem. Soc. 62, 3221-32 (1930). 17. Craxford, S. R., Trans. Faraday SOC.36,046-58 (1939). 18. Storch, H. H., I d . Eng. Chem. 37,34&51 (1945). 19. Fischer, F., Brennslofl-Chem. 11, 489-500 (1930). 20. Fischer, F., and Meyer, K., Brenmtofl-Chem. 12,225-8 (1931). 21. Fischer, F., and Koch, H., Btennstofl-Chem. 13,61-8 (1932). 22. Fischer, F., Pichler, H., and Reder, R., Branstoff-Chem. 13,421-8 (1932).
154
H. H.’ STORCII
Fischcr, F., Roelen, O., and Feisst,, W., fIrenmtof-Chern. 13,461-8 (1932). Hartner-Seherich, R., and Koch, IT., Drenr&o~-Chern.13,308-10 (1032). Fischcr, F., and Koch, H., Brennstog-Chem. 13, 428-34 (1932). Koch, H., and Horn, O., Brennstof-Chem. 13, 164-7 (1932). Anon., Irmt & Coal Trades Rev. 129, 542 (1034); Times (London), Nov. 19, 1934; Cheni. & M e t . Eng. 41, 666 (1934); Gas u. Wmserfach 77, 798 (1934); Fuel Economist 9, 303 (1034). 28. Fischcr, F.,and Koch, H., and Wiedeking, K., Brennstof-Chem. 16, 229-33 (1034). 29. Fischw, P.,Brennstof-Chem. 16, 1-11 (1935). 30. Thnu, A., Colliery Guardian 160, 335-7 (1935). 3 I . Koch, H., Cliickai~f71, 85-00 (1 035). 32. RIiirtin, F., Ind. Chemist 13, 320-6 (1937); Ocl, Kohlc, Tcer 13, 601-7 (1937); Chem. Fab~ilz12,233-40 (1!)39). 33. lluhrchamie A. G.; German Appl. No. R1049!39; 0.P.D. Rcport 412, May 15, 1945; British Patent No. 518334 (10.10); French Pnt,cnt No. 843305 (1939). 34. Hecltcl, II., nnd Roclcn, O., U. S,Patent No. 2219042 (1940). Schndc., 11. A., Form, E., nnd Aldrich, El. C., Hat. Petroleum N m s 37, RO22-4 (1945). 35. Hall, C. C., nnd Smith, S. L., J . SOC.Chem. Ind. London 06, 128-36 (1946). 36. Cnwlcy, C. AT., and Hall, C. C., Ann. Rev. Petrol. Tech. 6 , 361-83 (1940). 37. Pichler, H., The Synthesis of Hydrocarbons from Carbon Monoxide arid Hydrogen, [J. S.Ihircau of Mines Special Report (194i). 38. ICoch, H., and Hilberath, IT., Brenndof-Chem. 22, 135-9, 145-52 (1941); 23, 67-73 ( 1942). 39. Hilberath, F., Thesis, Kiel, Germnny (1940) ’79 pp. 40. voii Wcbcr, U., Angew. Chent.. 63, 607-10 (1939). 4 1 . Rcichl, E. H., U. S. Naval Technical Mission in Europc Report No. 248-45 (1945) 130 pp., PB 22841. 42. Atwell, H. V., and Schroeder, W.C., St*cinkohlen-BergwcrkRheinpreussen MoersMeerbeck, Hombcrg, Germany, CIOS File XXV-6 Item 30; P B 412 (1945). 43. Hall, C. C., Offices of Gukhoffnungshiittc A. G. Sterkrade, Ruhr, CIOS File XXVII-70 Item 30; P B 975 (1945). 44. Hnll, C. C., and Hacnscl, V., The Fischer-Tropsch Plant of Ruhrchemie A.G., Stcrkrade-Holten, Ruhr CIOS File XXVII-60 Item 30; P B 415 (1945). 45. Faragher, W. F., Home, M‘. A., Ilowes, D. A., Schindler, H., Chnffcc, C. C., West, H. I,., and Rosenfeld, L., Supplemental Rcport on Ruhrchemie A.G., SterkradeHolten (Oberhausen-Holtcn), Ruhr, CIOS File XXVII-96 Item 30 (1945). 46. Ncumann, R., and Schroeder, W. C., Fischer-Tropsch Plant of Hoesch Benzin, A.G. a t Dortmund, Germany, FIAT 239; P B 1279 (1945). 47. Horne, W. A., nnd Jones, J. P., Fischer-Tropsch Unit at Leipzig Gas Works, Leipzig, Germany, CIOS File XXVII-68 Item 30, PB 294 (1945). 48. Hall, C. C., Chemischc Werke Essener Steinkohle, A.G., Bergkamen near KnmenDortmund, Germany, Inspection of Fischer-Tropsch Plant April 19 and 20, 1945, CTOS File XXVII-54, Item 30, P B 413. 49. Hall, C. C., and Powcll, A. R., Plant of Klocknerwerke, A.G., Castrop-Rauxel, Gcirnmng, CIOS File XXV-7, Item 30, P B 286 (1945). 50. Atwell, H. V.,Poac*ll,A. R., and Storch, H. H., Fischer-Tropsch T.A.C. R.eport No. Sn hfc 1, P13 2051 (1945). 51. F M l , C. C., Craxford, S. R., nnd Gall, D., Interrogation of Dr. Otto 1l.oclc.n of Ruluchernie A.G. (1046), I3108 447, Item 30, reproduccd on T.O.M. Rccl22fi
23. 24. 25. 26. 27.
PROCESSES FOR SYhTHESIS OF HYDROCARBONS
155
52. Faragher, W. F ,and Horne, W. A., Intcrrogntion of Dr. Pier and Staff, I.G. Farbenindustrie A.G., Ludwigshafen and Oppnu, SuppIcment I1 Report of Investigations by the Fuels and Lubricants Tcnms, Bureau of Mincs Inf. Circ. 7376, August 1946. 53. Chaffce, C. C., Thompson, 0. F., King, J. G., Atwell, H. V., arid Jones, I. H., Metnllgesellschaft-Lurgi Frankfurt An hiairi, Germany, CIOS Filc XXXI-23, Item 30, PB 12624. 54. Technical Oil Mission Recl 134, Section lB, Itcm 21, Iteichl, E. H., sre rcfrrrncc (20). 55. Storch, H. H., Anderson, R. B., IIofer, L. J. E., Hawk, C. O., and Golumbic, N., Synthetic Liquid Fuels from Hydrogenation of Carbon hlonoxide, Part I, Bureau of Mines Technical Paper 709, i n press. 56. Murphree, E. V., Tyson, C. W.,Campbell, D. H., and Martin, H. Z., U. S. Patent No. 2360787 (1944). 57. Latta, J. E., and Walker, S. W., Chem. Eng. Progrrss 44, 173-176 (1948). 58. Ryan, P., Oil and Gas J. 43, No. 47, 264 (1!)45). 59. Kcith, P. C., Oil and Gas J . 46, No. 6, 102 (1946). 60. Johnson, E. A., U. S. Patent No. 2393009 (1946). 61. Keith, P. C., Chem. Eng. 53, No. 12, 101 (1946). 62. Murphree, E. V., Nat. Petroleum News 38, No. 41, 46 (1946). 63. Roberts, Jr., G., and Phinney, J . A., Oil and Gas J . 48, 72-3, 139-41 (1947). 64. Anon., Chem. Eng. News 24, No. 10,13i8 (1946); Nat. Petroleum News 98, No. 19, 51 (1946); Petroleim Refiner 26, No. 3, 142 (1946); Chem. Industries 68, No. 1, 134 (1946); OilandGas J. 46, No. 7, 58 (1846). 65. Crowell, J. H., nnd Bcnson, €1. E., Direct Evnporative Cooling in the Synthesis of Liquid Hydrocarbons froni Hydrogen and Carbon Monoxide, presented a t Am. Chem. SOC.Meeting, .4pril 1947, a t Atlantic City, N. J. 66. Weir, €1. RI., T h e OX0 Proccss for Alcohol hlnnufacture from Olefins, T.O.M. Report No. 6, PI3 2047 (1045). 67. Horne, W. A., Interrogation of Franz Fischer (1945), FIAT No. 90, P B 1253. 68. Faragher, W. F., and Horne, W. A., Interrogation of H. Pichlcr and I(. Ziegler a t Kaiser Wilhclm Institut fiir Bohlenforschung, FIAT No. 276, P B 1291. 69. Haensel, V., Kaiser Wilhelm Institut fiir Kohlcnforschung (1945), CIOS File XSV-1, Item 30, P B 284. 70. von Wangenheim, Ges. Abhandl. Kenntnis Kohle 8,227-33 (1929). 71. Fischer, F., and Bahr, H. A., Ges. Abhandl. Kenntnis Kohle 8, 255-68 (1929). 72. Fischcr, F., and Dilthey, P., Brenmtof-Chem. 9,24-30 (1928). 73. Hofer, L. J. E., The Preparation and Properties of Metal Carbides, U. S. Bureau of Mines Report of Investigations 3770 (July 1944). 74. Pichler, H., and Merkel, H., Chemical and hhgncto-Chcmical Study of Iron Catalyst, U. S. Bureau of Mines Special Report (1947). 75. Kummer, J. T., De Witt, T. W., and Emmett, P. H., Some Studies of thc FischerTropsch Synthesis using CI4, presented a t Am. Chein. Soc. Mccting, April 1947 a t Atlnntic City, N. J. 76. Fischer, F., and Kuster, H., Brennstof-Chem. 14,343 (1933). 77. Fischer, F., and Pichler, H., Brennstof-Chm. 12,365-72 (1931). 78. Myddlcton, W. W., and Walker, J. J., J . SOC.Chem. Znd., London 68, 121-4T (1036). 79. Tsuneoka, S., and Fujimura, K., J . SOC. Chern. Znd., Japan 37, Suppl. binding, 704-11 (1934). 80. Aicher, A., Myddleton, W. W., and Walker, J. J., J . SOC.Chem. Znd., London 64, 313-20T (1935).
156
H. €STORCH I.
81. Wellcr, S., Kinetics of Carbiding and Hydrocarbon Synthesis with Cobalt FischcrTropsch Catalysts, J. Am. C h m . SOC.68,2432-36 (1947). 82. Fischcr, F., and Pichler, H., Brenmtofl-Chem. 20,41-61 (1939). 83. Pichler, H., and Buffleb, H., Rrennstoff-Chem. 21,257-64 (1940). 84. Fitwhcr, F., and Pichler, H., Brenn&ff-Chem. 20,221-8 (1939). 85. Pichler, H., and Bufflcb, H., Brennstoff-Chem. 21, 273-80 (1940). 88. Trttnslations of German Documents on the Devclopmcnt of Iron Catalysts for the Fischer-Tropsch Process, Pnrt 1, by Mnx Lcva, U. S. Bureau of Mines Special Report (1946). Translations of Docs. PG-21559-NID, PG-215Sl-NID, PG-21574-NID, PG-21577-NID, nll from Technicnl Oil Mission Reel 101. 87. Hall, C. C., private communication. 88. Crnxford, S. R., Trans. Faruduy SOC.42, 576-80 (1946). 89. Craxford, S. R., and Ridenl, E. K., J. Chem. SOC.1939, 1601-10. DO. Wcllcr, S.,Hofcr, L. J . E., and Anderson, R. B., The Rolc of Bulk Cobnlt C*rrt)ide in the Fischer-Tropsch Synthcsis, J. Am. Chem. SOC.70, 799-801 (1948) 91. Hofcr, L. J. E., and Peebles, W. C., J. A m . Chem. Sor. 69,893-8 (1947).
The Catalytic Activation of Hydrogen D . D . ELEY Nash Lecturer i n Biophysical Chemistry. University of Bristol. Bristol. England
CONTENTS
.
I Introduction
.
.
.
.
.
.
.
.
.
.
I1. Experimental Studies of Chemisorbcd Hydrogen on Metals . 1. Methods . . . . . . . . . . . 2. The Accommodation Coefficient Method . . . . 3 . Heat of Adsorption Method . . . . . . . 4. Electrical Methods . . . . . . . 5. Other Methods . . . . . . . . . I11. Theoretical Studies of Chemisorbed Hydrogen on Metals . 1. Time of Adsorption . . . . . . . . 2. Nature of the Chemisorbed Film . . . . . 3. Mobilityof the Chemisorbed Film . . . . . 4 . Bond Typein the AdsorbedFilm . . . . . IV Production of Hydrogen Atoms by Hot Tungsten . . V Recombination of Hydrogen Atoms . . . . . VI . The Parahydrogen Conversion . . . . . . . 1 Early Work . . . . . . . . . . 2. The Paramagnetic Mechanism . . . . . . 3. The Chemical Mechanism . . . . . . . VII The Exchange of Atoms between Deuterium Gas and Hydrides VIII . Exchange and Hydrogenation of Unsaturated Hydrocarbons . . . . . . I X . The Fischer-Tropsch Synthesis . X . Dehydrogenation . . . . . . . . . . XI . Cyclization of n-Paraffis . . . . . . . . X I 1. The Hydrogen-Oxygen Reaction . . . . . . XI11. Biological Activation of Hydrogen . . . . . . XIV. General Conclusions . . . . . . . . . 1. Stabilityof the Chemisorbed Film . . . . . 2. Geometry of the Catalyst Surface . . . . 3. Energetics of the Transition State . . . . . . . . . . . . . References .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . .
. . . .
. . . .
.
.
.
.
.
.
.
.
.
. .
. .
. .
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
Page 157 158 158 160 161 163 164 166 166 166 168 170 172 173 174 174 175 176 181 184 187 188 189 190 191 193 193 193 194 195
I . INTRODUCTION The mechanism of the catalytic activation of hydrogen gas and of hydrogen in organic compounds is of importance for numerous industrial processes. and for the problem of biological oxidation . In addition. there is the quite general problem of the relation between catalyst constitution and 157
15s
D. D. ELEY
catalytic activity and the hydrogen molecule, because of its simplicity, forms a very suitable substrate for investigations of this kind. The successive discoveries of atomic hydrogen, parahydrogen, and deuterium have given a great impetus to studies of catalysis, and the present article does not attempt to review all the work in this field. We shall stress mainly the work of Rideal and his school at Cambridge University, the results of other schools being treated in less detail. The adsorption investigations of the late J. K. Roberts, using brilliant new techniques, has led to the view that many gases are chemisorbed by clean metal surfaces a t room temperature in a rapid and irreversible fashion. The chemisorbed layer of hydrogen on tungsten, for example, only starts to evaporate into a good vacuum at 500” C., and for oxygen the temperature is much higher. This work, together with kinetic investigations of certain catalytic reactions (particularly the parahydrogen conversion and the exchange reaction between deuterium and ethylene) led Rideal (1) and his collaborators t o the view that a catalytic reaction may occur between one gas adsorbed in a primary chemisorbed layer on the catalyst, and a second gas adsorbed above it, either in a true van der Waals’ layer or at gaps in the chemisorbed layer. Rideal has summarized his contributions in his Sabatier lecture to the Society of Chemical Industry (2). This view is a distinct devclopment in the theory of Catalysis, and may be distinguished clearly from the Langmuir-Hinshelwood picture (3), in which two reacting gases compete for the chemisorbed layer, reaction occurring between two adjacent atoms or radicals. This latter picture, requiring, as it does, a close balance of two free energies of chemisorption, is indeed unlikely a t low temperaturea. I n ndvancing this new picture of catalysis, the intention is to supplement rather than supplant the old picture, which is probably valid for many systems, particularly a t relatively high temperatures. A decision between the two mechanisms may in certain cases be based on a measurement of heats of chemisorption, where the chemisorption concerned is relevant to the catalysis. In the present author’s opinion this is only likely to be SO for ‘Lclca~i” plane metal surfaces, where there is good reason to believe that the larger part, of the surface is catalytically active. Until the details of adsorption m d catdysis have been worked out on such surfaces, the interpretation of results on more complicated catalysts, such as promuted oxides, is bound to involve an element of uncertainty.
1 I.
EXPERIMENTAL STUDIES OF CHEMISORnED
HYDROGEN ON METALS
1 . Methods J. K. Roberts (4, 5 ) initiated important work in this field. In his first method the chemisorbcd film was detected by it,s cfycct on the accommoda-
CATALYTIC ACTIVATION OF HYDROQEN
159
tion coefficient of the metal. In his second method the heat evolved on adsorption of the gas was measured. R. C. L. Bosworth and E. K. Rideal (6, 7) detected the adsorbed film by its effect on the electrical contact potential of the metal. I n all these studies a tungsten wire was employed, because its high melting point allows it to be heated in vacuo to temperatures of 2500' K., when all surface contamination is removed. The surface of the wire may be taken as a roughness factor p times its geometrical area, where p may be between 1 and 2. It has also been found possible to prepare clean tungsten surfaces of much larger area, e.g., 400 cm.2 by evaporation from a wire (8). This method has the advantage that larger volumes of gas are adsorbed, but the disadvantage that the true area cannot be obtained
I
To hydro cn oi~pply
t
an puinp
To pump
t
t
B
I LJ
To Macleod &uge
t
B
FIG.1. Measurement of accommodation coefficients.*
from the geometrical area. The method has also been used to obtain clean surfaces of other metals, such as nickel, with areas of 10,000 cm.2 (9). * Figures 1 , 2. 3. 4, 7, and 8 are reproduced from J. K. Roberts "Some Problem in Adnorption
"
Canibridge Univeraity Press, London, 1939. Grateful acknowledgment i~ recorded to the author and pu&lisher concerned.
100
D. D. ELEY
8. The Accommodation C o e f i e n t Method (4) Neon gaa at 0.1-mm. mercury pressure was circulated over a tungsten wire 18 cm.XO.0066 cm. sealed in tube A. Charcoal traps B cooled in liquid air protected the wire from oxygen, and a liquid air trap was inserted between the wire and the McLeod gauge. A standard current could be passed through the wire, raising its temperature about 20" above that of a bath surrounding A. The resistance of the wire, and hence its temperature T2,waa determined by a Wheatstone bridge. An additional circuit allowed the wire to be "flashed" for a short time a t 2500" K. to remove adsorbed films. After circulating neon for some time, the wire waa flashed and then switched over to the standard current, values of the wire resistance being taken a t intervals of time from zero time (the time at which the flashing current was switched off). If TIbe the temperature of the gas in tube A, q, the heat loss frpm the wire in cal. cm;' set.-', is given by Knudsen's equation, Q = 1.74
x 10-4
FaPGI(Ta- T,)
where a is the accommodation coefficient, M the molecular weight, and p the pressure in dynes crn-'. of the neon. Fig. 2 shows a plot of a against
FIQ.2. Effect of adsorbed hydrogen on the accommodation coefficient. The arrow marke the time at which the flashing current w a cut ~ off. time, showing a very slight drift due to adsorption of traces of impurity. Extrapolation back to zero time gives a value of a characteristic of the clcan metal surface.
CATALYTIC ACTIVATION O F HYDROGEN
161
A trace of hydrogen was now admitted to the wire via the charcoal trap, sufficient to give a pressure of 4x10" mm. mercury. An immediate rise in a was found, a steady value being reached which was the same even if a ten times larger dose of hydrogen was admitted. It seems clear that a complete monolayer of hydrogen has formed on the wire, the adsorption process being rapid even at 79" K. A trace of admitted air was completely taken up by the charcoal traps and the effect observed was therefore not due to impurities in the hydrogen. A similar method has been used to detect oxygen films, and it was possible to distinguish three distinct types of film in this case (4,10, 11). 3. Heat of Adsorption Method (4) The apparatus comprised a tube, containing a fine tungsten wire, to which was attached a pirani gage calibrated for hydrogen, and a gas pipette. As before, the apparatus was pumped to a hard vacuum and the wire flashed
Time
FIG.3. A plot of the resistance of the wire (galvanometerreading) a g h t time. The arrow indicates the time at which hydrogen waa admitted.
As the flashing current waa cut off, the wire was switched into a sensitive Wheatstone bridge in which a Paschen galvanometer was used. The decrease of the wire resistance with time aa the wire cooled down is shown in Fig. 3.
D. D. ELEY
162
On admission of a charge of hydrogen to the mire from the pipette the wirc temperature rose abruptly. Successive charges were admitted until the pirani gage indicated the appearance of hydrogen in the gas phase, and thus the saturation of the adsorbed film on the tungsten. A wire of superficial area 0.55 was found to chemisorb 4.4X1OX4molecules.
I
101 0
I
I
I
I
t
I
0.5 n,jY.X x
I
I
1 0
10''
r ~ = i i i c : i nnuiiibrr
of filled pl;iccs
Fro. 4. Heat of adsorption per mole hydrogen 88 a funotion of surface covered.
From the dimexisions and heat capacity of the wire the heah of adsorption could be calculated. Fig. 4 shows how this decreases from 45 kcal./mole hydrogen for a bare surface to 18 kcal./mole hydrogen for a nearly saturated surface. With oxygen gas (4) a primary layer had a mean heat of adsorption of 139 kcal. and contained 4X1Ol4 molecules. A secondary film was also detected with a heat of 48 kcal./mole oxygen which contained O.9X10l4 molecules. The method has also been used to detect adsorbed nitrogen (12).
CATALYTIC ACTIVATION O F HYDROGEN
163
4 . Electrical Methods J. H. de Boer (13) has summarized the early work of Langmuir on the effects of adsorbed films on thermionic emission from metal surfaces. This technique cannot give information about adsorbed films in the temperature range of interest in catalysis, say below 500" C. In this range one may detect adsorbed gases by a study of the photoelectric work function, or the contact potential. The first method offers difficulties, since with hydrogen and other gases the threshold is in the far ultraviolet. Some results for adsorbed hydrogen have been reported by Suhrmann and Csech (14) and bypunsky (15). The second method is relatively simple to operate and has given a great deal of information in the hands of Bosworth and Rideal (6, 7). It was originally developed by Langmuir and Kingdon (16), who laid down conditions for its successful use. The reaction tube contains two tungsten wires, the emitter and collector, spotwelded to leads so that they cross a t right angles with a separating gap of about 1 mm. An example of such a tube, in which the collector is a wire long enough to function as a catalyst filament, is shown in Fig. 5. For adsorption work only, a shorter collector as used by Bosworth is more suitable, since it heats up less under the influence of radiation from the emitter. The tube is connected via a liquid air trap to a vacuum apparatus. The best results are obtained when a film of sodium is distilled over the inside of the tube, to act as a trap for oxygen. The whole reaction tube is normally immersed in liquid air. The filaments may EIQ.5. be heated electrically by separate circuits, and an The measurement additional circuit allows a potential to be applied ofcontactpotential. between the two wires and the associated thermionic Reproduced from current to be measured. Eley and Rideal, I n his work on hydrogen films (7) Bosworth first P ~ cRoy.Soc.Lon. cleaned both filaments by flashing in vacuo. The col- d o n 429 (1941). lector was allowed to cool, the emitter adjusted to a standard temperature, e.g., 2000" K., and the thermionic current-polarizing voltage curve taken. This is curve A in Fig. 6. The emitter was now switched off, a dose of hydrogen admitted, a few seconds allowed for adsorption, and the apparatus evacuated. The emitter was switched on and the current-voltage curve again taken, curve B being obtained. The straight part of B is parallel to the straight part of A, but shifted to more
164
D. D. ELEY
positive potentids by an amount A V , which is minus the W-WR contact poten tial. Bosworth obtained values for W-WH of - 1.04V1W-WD, - 1.02V (7), W-WO, -1.76V1 W-WN, -1.38V (17).
VOLTAGE
Fra. 0. Current-voltnge curves for W-W(A) and W-WH(B) surfaces.
We may probably neglect the mutual depolarization of the adsorbed dipoles for hydrogen, nitrogen, and oxygen, when the dipole moment may be calculated by A\’ = ~ T ~ L C , ,
cd being the surface concentration of dipoles, which for a saturated surface is usually taken as equal to the number of adsorption sites per square centimeter. I t may, however, only be a definite fraction of this number. For hydrogen p = -0.40 and for oxygen p = -0.66D,the sign indicating that the outer layer, hydrogen or oxygen, is negative. The method was used (17) to follow rates of evaporation of adsorbed monolayers of these gases.
6 . Other Methods
I t was reported in 1929 (18, 19) that hydrogen adsorbed on metals gave a characteristic electron diffraction pattern, but the work has not been folloived up. I n the ordinary electron diffraction apparatus one might
CATALYTIC ACTIVATION OF HYDROQEN
165
expect difficulties to arise from adsorption of impurities and desorption of hydrogen caused by impact of the high energy electrons. The most important use of this instrument to date would seem to be the determination of lattice orientation and spacing of the catalyst. By bombardment with electrons adsorbed films of hydrogen have been found to possess an ionization potential of 13.4V (20, 21), but the effect is a complex one, and this value does not necessarily establish the presence of atomic hydrogen on the catalyst (22). Eley (8) has shown how substantially clean tungsten surfaces of large area may be prepared by evaporation from a wire on to the walls of the reaction vessel. With a good high vacuum technique such films may be preserved free from adsorbed oxygen, and therefore unpoisoned, for a week or longer. Admission of hydrogen to such a film resulted in an immediate uptake of a quantity of gas measurable on a McLeod gage. The adsorption was irreversible and the gas could not be removed by pumping the film at 293” K. Since the true area of the film was unknown, it had to be determined by calculation from the observed uptake of’ hydrogen, assuming 1 hydrogen atom per tungsten atom in the surface. In other words, Roberts’ results must necessarily be assumed to obtain the film area. The method, however, has confirmed Roberts’ results on the speed of formation and stable nature of the chemisorbed film. Beeck et al. (9,33d) have given similar confirmation for films of nickel and other metals. They have made the important observation that evaporation of the metal in the presence of inert gas leads to oriented films. It seems clear that the films are microcrystalline, the crystalline size depending on the conditions of evaporation (23a). Frankenburger (23b) and coworkers have evaporated metal wires in the presence of gases. This calls to mind the classical study of Langmuir (23c) of the reaction between tungsten and nitrogen, which he showed to occur in the gas phase. Frankenburg found the reaction between iron and hydrogen to occur on the walls of the vessel. The conditions are complex but the method clearly merits further study. Films thus offer a means of preparing clean surfaces of relatively low melting metals. Once a strongly chemisorbed film has been formed, however, it cannot be removed. Where this is desirable one can sometimes use the metal as a coat on a tungsten wire. In this way Bosworth (24) determined the Ni-NiO contact potential as - 1.4V. The cIassicaI method of study of hydrogen adsorption is, of course, to follow the uptake of gas by the powdered adsorbent by means of a manometer. Difficulties associated with this method concern the “cleaning” of the metal surface before adsorption, and additional gas uptake due to solution in the metal. These difficulties, realized in 1932, are still not
166
D. D. ELEY
resolved, and the reader is referred to the Faraday Society Discussion of 1932 for a full account of the method and its results. Frankenburg (23d) has recently published a careful study of the system hydrogen gas-tungsten powder. 111. THEORETICAL STUDIESOF CHEMISORBED HYDROGEN ON METALS 1. Time of Adsorption The known cases of chemisorption of gases on clean metal surfaces occur so rapidly that they must possess a negligible activation energy. They are quite different from the d 0 W adsorptions, called “activated adsorptions,” described by H. s. Taylor (25). Restricting our attention to tungsten, it has been found that tungsten powder, outgassed at 750” C., shows an “activated adsorption” of hydrogen (26) and nitrogen (27). The affinity of oxygen for a tungsten surface is such that we cannot expect to remove it by pumping the powder for a few hours at even 900” C.* In this connection Roberts (4) discovered that a tungsten wire coveredwith chemisorbed oxygen will show a slow uptake of hydrogen, which he attributed to a diffusion of oxygen into the interior, thus setting free sites for adsorption. This gives a plausible account of activated adsorption in this case. In this particular case there would also seem a possibility that the oxygen is reduced to water (28). Beeck el al. (9) have noted that a nickel surface will lose oxygen to the interior on standing, even in the absence of hydrogen. For metals, at least, we conclude that “activated adsorption’’ may be connected with the establishment of chemisorbed layers in the case of impure catalysts, but not with chemisorption on a clean metal surface, as put forward by Taylor (25). It follows, therefore that “activated” adsorption of hydrogen may be of significance for the initial process of activation of technical catalysts, but of less siGificance for catalytic reactions occurring in the established chemisorbed films. Further experiments on other systems, such as tungsten-methane, are required before we can generalize this picture. Furthermore, the author would restrict his opinion to metals, as the more complex metal oxide systems require separate consideration. 2. Nature of the Chemisorbed Film
(4, 6 )
In section 11, 3 it was stated that a tungsten wire of apparent area 0.55 cm.2adsorbed at saturation 4.4X 1014molecules. Assuming a tungsten
wire exposes 100 and 110 planes equally (29,30), the surface would contain 6.65X1Ol4 p tungsten atoms, the roughness factor p reasonably lying between 1and 2. The film may therefore be either (a) made up of hydrogen *Where a bulb is heated, one must also bear in mind diffusion of gnses from the atmosphere through the walk
CATALYTIC ACTIVATION O F HYDROGEN
167
atoms, 1atom being held by each surface tungsten atom or site, or (b) made up of hydrogen molecules, 1 hydrogen molecule being held per site, and excluding the occupation of its four nearest neighbor sites. With Pauling’s (31) values for the radii concerned, this would be so for the 110 plane, but on the 100plane the molecules would probably pack with 1 molecule per site. While structure (b) is not excluded by the evidence, structure (a) is more likely, since the high heat of adsorption would suggest the formation of chemical bonds. On present valency theory, it is difficult to see how molecular hydroge-n could form a bond of the strength observed. For oxygen there are at least three possibilities for the primary film, and while Roberts originally considered (4) it to be atomic, he later considered (32) it to be molecular, of type (b). In this case the ?r electrons are available for bonding without disruption of the molecule into atoms. No evidence as to the nature of adsorbed films of other important gases, such as CO, Nz, CHI, and C2H4is yet available, apart from the relative adsorption values of Beeck et al. (9) on nickel films. These authors give as the number of molecules adsorbed per site CO, 1; HI, %; Nz, %; CzH4,%; 0 2 , 2. The oxygen result they think is due to diffusion of oxygen under the surface. Ethylene, they suggest, is adsorbed with 1 hydrogen atom per site, but this method would scarcely give a strong bonding. If we allow for an ethylene molecule excluding the occupation of neighboring sites, as in Roberts’ case (b) above, we find the observed result may arise in a number of ways, e.g., (1) Two point adsorption at the double bond. Twigg and Rideal’s (33a) considerations suggest that on the 100 plane neighbor interactions might just be large enough to result in 1 in 4 occupation. (2) One point adsorption of ethylene, either flat on the surface, or vertically as CH3CH= or C H A H - fragments. It seems likely that other physical data, such as contact potentials, will be required to decide the mode of adsorption in cases like ethylene. As we shall see, data on the catalyzed exchange reactions of ethylene are most easily interpreted on the basis of two-point adsorption at the double bond. This is indeed what we should expect from the reactive character of the mobile ?r electrons. W. G. Frankenburg (23d) has examined the adsorption of hydrogen on tungsten powder. From the strong decrease of heat of adsorption with surface covering, he has concluded that the first hydrogen is adsorbed as atoms, but subsequently the gas is taken up as molecules in “varying degrees of strain.” He attempts to reconcile his results with Roberts’ heats of adsorption on a wire, by assuming that Roberts’ saturation value corresponds to 10-20% surface covering of the wire. This would require a roughness factor for the wire of 4.5, which is very unlikely, the value for a
168
I). D. ELEY
tungsten wire similarly treated being only 1.35 (33b). The roughness factor may be supposed to lie somewhere between 1and 2 (5). In a future comparison of this kind, further work on the roughness factor is necessary, and at the same time, consideration must ge given to possible solubility effects in the uptake of gas by the powder, which would also lead to the kind of discrepancy found by Frankenburg. In a recent paper, G. Halsey and H. S. Taylor (33c) discuss Frankenburg’s results in the light of R. H. Fowler’s treatment of adsorption. The data require, in their view, the hypothesis of a nonuniform surface. This may well be so for such a powder, but they are wrong to conclude that “heterogeneity is especially significant in the case of tungsten metal which has hitherto been treated as an essentially uniform surface.” A completely aged tungsten wire may still be expected to be essentially uniform, whatever may be the cam for the surface of the metal powder. A paper by 0. Beeck (33d) has only recently come to the author’s notice. Working with evaporated films of metal of area of the order of 10,000 cm.2 Beeck has confirmed all Roberts’ results on the speed, extent, and heat of adsorption as a function of surface covered, for the adsorption of hydrogen on tungsten. Nickel shows a similar behavior. This makes extremely likely the view that adsorption on tungsten powder is complicated in some fashion, if not by solution, as suggested above, then by some kind of physical or chemical heterogeneity of the surface, as in the Halsey and Taylor picture. 3. Mobility of the Chemismbed Film Lennard-Jones (34) showed theoretically that lateral diffusion ‘of adsorbed atoms would require an activation energy. The available evidence suggests a value for this one-quarter to one-third the evaporation energy of the hydrogen as atoms, i.e., about 20 kcal./g. atom. With this value we should not expect much lateral diffusion of atoms at temperatures equal or less than room temperature. Roberts’ views on this important problem rest on his analysis of his heat of adsorption data (Fig. 4). He considers a quadratic lattice, and assumes the only interaction energy is a fixed energy V between nearest neighbors. This corresponds to a repulsive force, partly at least dipolar in nature. He distinguishes two extreme cases ( 5 ) :
(a) The hydrogen molecules dissociate into atoms, which are immobile and remain on their original sites. (b) The 6lm is formed as before, but the atoms are supposed mobile and therefore move into positions of lowest energy.
169
CATALYTIC ACTIYATION OF HYDROGEN
If 0 be the fraction of occupied sites, assumption (a) leads to the linear relationship found in Fig. 4. q = Po - zev,
where qo is the heat of adsorption on the bare surface and Z,the number of nearest neighbors, is 4. On assumption (b), however, up to 0-0.5, the atoms may arrange themselves so that no two atoms are nearest neighbors. This may be seen by reference to Fig. 7.
Fro. 7. Mobile a t o m on a quadratio lattice.
Thus, q=qo up to 0=0.5, at which point it abruptly falls to qo-4V, as shown in curve a, Fig. 8. By taking account of Boltzmann's distribution, the corners of the curve are rounded to an extent depending on temperature, or more precisely, on 7 = e-'IkT. WaDg (35) has given a full treatment, obtaining the equation q-qo
V
- -E 2
L1
-{1-4(1
I -228 -q)e(l-e)t
}
1
Curves for various values of 9 are given in Fig. 8. These considerations, based upon a fixed interaction energy V , thus ilm at room temperature. However, definitely favor an immobile hydrogen f Miller and Roberts (36)refined the theory so as to allow for an alteration in V as the distance between adsorbed atoms changes because of thermal vibrations. Cases (a) and (b) now both give a very flat sigmoid, almost
D. D. ELEY
170
linear, curve, and so the mobility of the hydrogen film at least at293"K., remains an open question. We may, however, suppose that a t very low temperatures, e.g., 79" I<., the hydrogen film is immobile. Roberts (4)has shown that the dissociative
Fro. 8. Variation of heat of adsorption with fraction of surface covered. Curve (a), q=e-"CkT=O; (b), q=3.6X10-8; (c), q=8.2X10 *; (d), q=0.308; (el, q = 1. The scale on the left is for the heat of adsorption g when there is no dissociation and the one on the right is for the heat pcr molecule p2 when thew is dissociation. The subscript, 0 refers to e= 0, and V is the intcrnction energy betwccn particles adsorbed on neighboring sitcs. The crtlculat,ion.sarc for
2=4.
adsorption of diatomic molecules to form an immobile atomic film results, on purely statistical grounds, in the complete film containing 87, empty sites or gaps. He also showed how gaps may arise in either mobile or immobile films, where the adsorbed molecule is so large that it excludes occupation of neighboring sites (3). He suggested that such empty sites might play a part in catalytic action, and catalytic mechanisms based on the assumed presence of gaps have been developed by Rideal (1).
4. Bond Type in the Adsorbed Film Taking 103 kcal. as thc dissociation energy of the hydrogen molecule, , Roberts' adsorption heat of 45 kcal. gives a W-H bond strength (dilute film) of (103+45)/2=74 kcal./g. atom. Bosworth gives the dipole moment as 0.4 Debyes, the hydrogen being negative. Calculations of the dissociative adsorption of hydrogen have been made
CATALYTIC ACTIVATION O F HYDROGEN
171
on carbon (37) and nickel (38), using the Polanyi-Eyring method (39). These calculations treat the surface atoms of the adsorbent as free atoms, and ignore the essential character of the metal electrons, which are not localized on definite atoms but exist in energy bands common to the whole crystal. The calculations do, however, indicate one way in which lattice spacing may affect the energy curve for adsorption. We may suppose that it is essential to use a proper model for the metal. The adsorption of cesium on tungsten, a particularly simple case since the cesium ionizes completely on the metal surface, has been treated in this way most successfully by Langmuir (40) and de Boer (41). Catalysis requires a similar treatment for electronegative gases on metals, where the bond formed is largely homopolar, and not ionic as in the case of cesium. The problem has been treated in an illuminating fashion by LennardJones (34). He shows how an adsorbed atom at a metal surface will perturb the periodic potential field (due to the regular lattice of metal ions) in the neighborhood of the metal surface. This will lead to a local energy level in the metal surface next to the adsorbed atom and an electron in this level in favorable conditions will “exchange” with that in the adsorbed atom, leading to an essentially homopolar bond. Attempts have also been made to treat the surface energy levels of metals in the absence of adsorbed gases, by Tamm (42) and later workers. References to this work will be found in a paper by Shockley (43). Pollard (44) has endeavored to combine Tamm’s considerations with those of Lennard-Jones. He has calculated the adsorption potential energy curve for hydrogen on a metal, finding an activation energy for adsorption similar to those discussed by Taylor (25). He considers that a one-electron bond will be formed (as in Hz’), for the case where the energy levels are equal in the adsorbed atom and the surface metallic state. The general approach adopted is of great value. The present author considers it essential that in theoretical considerations of catalysis the adsorption forces should be considered as arising from the metallic electrons, as indeed Polanyi (45) pointed out many years ago. For ethylene, for example, he would regard the ?r electrons in the adsorbed molecule as forming part of the electron gas of the metal. Such a concept does not involve the positive end of the dipole on the adsorbed atom. There is no conflict between the negative charge on adsorbed hydrogen on tungsten, and the existence of dissolved hydrogen as protons within a metal lattice (46). In each case the electrons lie in a metallic level, though in the adsorption case this level may be common only to the surface layer and not the whole metal. As Franck (47) observed, a dissolved proton must be surrounded by an electron swarm, otherwise the high ionization potential of hydrogen atoms precludes an explanation of the heat of solution observed. This notion applies equally to adsorbed hydrogen.
172
D. D. ELEY
IV. PRODUCTION OF HYDROGEN - 4 ~ 0 BY ~ 8HOTTUNGSTEN I. Langmuir (48) showcd that if hydrogen gas were introduced into a bulb containing a heated tungsten filament, and the walls cooled in liquid air, a continuous decrease in pressure was observed. He supposed hydrogen molecules dissociated on the filament to give chemisorbed atoms, which evaporated and were trapped on the cold walls. 11, + 2W +2WH +2W + 2H G. Bryce (49) reexamined the reaction, using a layer of molybdenum oxide on the walls of the vessel to trap the atoms. This method of trapping he found more efficient than the previous one. Over the temperature range 1148 to 1420" K. and pressure p 2.4XlO-' to 37X10-3 mm. mercury, n, the number of hydrogen atoms produced per cm.2per sec. is Roberts and Bryce (50) consider two processes: (a) The evaporation of atoms from the surface and the reverse proceM of adsorption of atoms from the gas. (b) A hydrogen molecule strikes an empty site on the metal surface, one atom being adsorbed and the other returning to the gas phase. The reverse process is a gas atom striking an adsorbed atom and forming a molecule, which evaporates. For equilibrium, 0 the fraction of surface covered by atoms, for a hydrogen molecular pressure pn, is given by o 1-0
where A is a constant given by statistical mechanics. Equivalent values of A may be obtained by applying the principle of detailed balancing t o processes (a) and (b). Process (a)
Process (b)
K is the equilibrium constant in atmospheres for H G 2 H in the gas phase. mlis the m a of a hydrogen atom, ms that of a molecule, a is the probability
that an atom striking ti bare site is adsorbed, v the probability that an atom will evaporate, 8 the probability that a molecule striking a surface will split up into an adsorbed and a gaseous atom, and y the probability of the reverse step to the last named. Process (a) then yields for the rate of production of atoms, u,
CATALYTIC ACTIVATION OF HYDROGEN
173
Process (b) yields u-
pz*
(2rmzk1')1
(1
- 0)
Thus a kinetic equation of the correct order may arise either from process (a) with a low surface covering e, or from process (b) on a nearly saturated surface (0-1). In each case the main effect of temperature is exerted through the term K' . Assuming the observed velocity at 1420"K., the authors calculated the velocity a t the other temperatures using Giauque's data (51) and a value for the dissociation energy of hydrogen of 101 kcal. The results are shown in Table I, indicating a good agreement with experiment. TABLE I Velocity of Production of H atom, v, in atomcr per square centimeter per second at p~ = 29.~iXlO-~ mm. mercury 1)
x10 -1'
T"K,
Obs.
Cdc.
1420 1338 1243 1148
316 115 32
(316) 103 22 4
7
Bosworth (7)determined the contact potential of a tungsten filament under the above conditions, deducing that 0 lay in the range 0.6-0.9. It was therefore concluded that for the conditions concerned, mechanism (b) waa operative. The experimental conditions departed from the equilibrium conditions assumed in the above theory, but we may suppose the resulting error is small. Other workers (52) have attempted to reconcile process (a) with Bosworth's result by assuming the reaction occurs on "active centres.'' As we shall see, a vaiiation in activity over a tungsten surface is detectable in surface exchange reactions at 77" K., but the variation is so slight that it is undetectable at 193°K. At 1100°K. there is therefore every reason to believe the tungsten surface is uniform in its activity.
V. RECOMBINATION OF HYDROGEN ATOMS Bonhoeffer (53) used a thermometer bulb coated with a thin film of the catalyst. On exposure to a stream of hydrogen gas containing atoms, the thermometer reached a temperature which depended on the efficiency of
174
D. D. ELEY
the catalyst. The metals in order of dccrcasing efficiency were palladium, platinum, silver, tungsten, iron, copper, calcium, lead, and mercury, which is the order of their hydrogen overvoltttgc. He concluded that the atoms recombined in the adsorbed layer to evaporate as molecules. Roginsky and Schechter (54, 55) found that the recombination of atoms possesses a smdl, though definite, activation energy. This, together with the specificity exemplified in the series above, shows that the catalyst is not functioning as a mere third body for the removal of the energy of reaction. These authors are in favor, for their experimental conditions, of a mechanism involving a gas atom striking an adsorbed atom and evaporating as a molecule. This fits in with Bryce and Roberts' mechanism for the production of hydrogen atoms. VI. THEPARAHYDROGEN CONVERSION 1 . Early Work
In their classical paper Bonhoeffer and Harterk (56) distinguished between the catalysts charcoal and platinum black, both of which are active in causing the conversion reaction parahydrogen G orthohydrogen. Charcoal is most efficient at low temperatures, and is therefore used t o prepare parahydrogen a t 77" K. or lower temperatures. Platinum black is inefficient a t low temperatures, but its efficiency increases with temperature. I t was subsequently possible (57) to distinguish two mechanisms, which we may call the paramagnetic and chemical mechanisms. Active charcoal wa8 considered to be a paramagnetic catalyst, that is, the hydrogen was converted by the magnetic field of the unsaturated surface atoms of the catalyst. This mechanism was established for two classes of homogeneous catalyst, paramagnetic gases and paramagnetic ions in solution (58), and treated theoretically by Wigner (59). According to Wigner, the transition probability is little affected by temperature, so that the conversion rate depends mainly on the concentration of hydrogen molecules in the van der Waals layer, which increases with decrertse of temperature, t o give the result, observed. Paramagnetic metal oxides were also found to be effective in this way, but diamagnetic oxides less so (60). ParticuIarly convincing is the result that oxygen adsorbed on charcoal at low temperatures, and therefore presumably in the form of paramagnetic molecules, promotes the conversion. If the oxygen is adsorbed a t higher temperatures it exerts a poisoning effect, attributed to adsorption as atoms (57). The chemical mechanism (61, 62) was supposed to involve the d i s m ciative chemisorption of hydrogen molecules, in a loosely bound fashion. In this mechanism, the hydrogen, dissociated into atoms on the surface of the cataIyst, recombined to give normal hydrogen, i.e., a mixture in which the
CATALYTIC ACTIVATION O F HYDROGEN
175
ortho/para ratio had the equilibrium value for the temperature of experiment. pHz
+ 2hI e2hIH
2M
+ TLH,
A parallelism was drawn with Taylor's notions about activated adsorption. The reaction Hz+Dz+2HD on nickel (63) was found t o possess similar kinetics to the conversion, and the same mechanism was proposed for this reaction. On the other hand, this reaction did not go on charcoal at low temperatures (64,65), thus confirming the paramagnetic mechanism in this case. The work up to 1935 is admirably summarized in the book by A. Farkas (66), and we restrict ourselves to a consideration of more recent work. 2.
The Paramagnetic Mechanism
The original explanation still holds good. It has, however, occurred to the author that the normal method of comparing the conversion with the H2+D2 reaction does not allow a completely unequivocal decision as t o whether the conversion is paramagnetic or not, a t temperatures as low as 77°K. Chromium oxide a t this temperature shows a very rapid conversion, but a relatively slow Ha+Dz reaction (64). A difference of activation energy of 1.2 kcal. such as may arise from zero point energy effects (63) would. cause the H2+D2reaction to go more than a thousand times more slowly than the conversion a t this temperature, if the latter involved only the chemical mechanism. While the paramagnetic substances hemin, hematin, and copper phthalocyanine catalyze the conversion a t room temperature, the diamagnetic hematoporphyrin and metal-free phthalocyanine are inactive (67). No catalysis was found for the H2+DZ reaction. Since the surface area of the hemin crystals could be estimated, the collision efficiency of the conversion could be calculated. The value of 10-'oso determined is in reasonable agreement with Wigner's theory (59). Burstein (67) has found that very pure charcoals stringently outgassed at 900" C. become very active catalysts for the Hg+D2 reaction at liquid air temperature. Her conclusion was that the conversion involves an exchange mechanism similar t o that reviewed in the next section. One is much inclined to suppose that the paramagnetic conversions observed on the earlier charcoals were due t o adsorbed impurities rather than t o surface free valencies. We might expect that the surface was saturated with chemisorbed oxygen, with some normal oxygen molecules held firmly over gaps in this layer and active as paramagnetic centers. A temperature of 900"C. is required to free a tungsten surface slowly from oxygen, and a similar result may hold for charcoal. Thus, Burstein's stringent outgassing may
176
D. D. ELEY
have given a clean surface which could chemisorb hydrogen an'd convert parahydrogen by the mechanism below. Turkevich and Selwood (68) have examined the conversion on the solid free radical a,&-diphenyl-8-picrylhydrazyl. This substance did not show a very strong van der Waals' adsorption of hydrogen, and its efficiency a6 a catalyst waa much increased by mixing it with zinc oxide, which by itself was a powerful adsorbent but a weak catalyst. 3. The Chemical Mechanism a. The Surjaee Interchange Reaction. The original term "high temperature mechanism" is a misnomer now we know that the conversion can go through this mechanism at 77" K. on tungsten (8, 28), charcoal (67) and chromium oxide (64). A. Farkas (69) found the parahydrogen conversion was catalyzed by a tungsten wire and described his results in terms of the dissociation and recombination of hydrogen in a loosely bound chemisorbed layer. Roberts' results on the stability of the chemisorbed film showed that this mechanism was unlikely to hold for the specific catalyst and conditions concerned (70). The rate of combination of chemisorbed atoms at room temperature is immeasurably slow and cannot account for the extremely rapid conversion observed. In reply, Farkas suggested (71) that most of the chemisorbed hydrogen was inactive, the conversion proceeding by his mechanism for a few active spots, not detectable in Roberts' work. The matter waa subjected to experimental test by Eley and %deal (8, 28). While Farkas (69) had shown the need to flaah a tungsten wire at 2500" K. to make it catalytically active, it seemed desirable, in view of the known activity of bulk oxides, to establish the abeence or otherwise of surjace oxide from the active wire by means of an independent physical method. A stringent high vacuum technique was used, and the contact potential method adapted to measuring the concentration of adsorbed oxygen on the catalyst wire. The reaction vessel, with its long catalyst (collector) filament, and its short emitter filament, has been shown in Fig. 6. With the long collector it was conveniently found that the heat radiation from the emitter generated a sufficient temperature to evaporate hydrogen from a mixed surface film of hydrogen and oxygen on the collector. "he temperature reached was insufficient to evaporate the oxygen, the surface concentration, c,, of which was calculated from the observed contact potential AV by A V = 27rpc,, where p = 0.66D. The procedure nw to clean both filaments, form an oxygen film on them, and then to flash thc oxygen off the emitter. The catalyst filament was then heated until itc contact potential indicated the required c, value. Both filaments were ther brought to bath temperature (77" K.), parahydrogen admitted, the catalyst
CATALYTIC ACTIVATION O F HYDROGEN
177
wire switched on to give its standard operating temperature of 142" K., and the parahydrogen conversion followed as a function of time. As in all other cases, a first order constant k described the course of the reaction. At the end of the experiment, the tube was evacuated and the contact potential, and thus cs, checked. The cont,act potential for a full oxygen film, primary film plus oxygen on the gaps, tvaa found to be -1.74V, and a value of - 1.22Vwas attributed to theprimaryfilm. Amere removal of oxygen from gaps did not allow the conversion to proceed, but the activity of the surface increased strongly as the primary chemisorbed oxygen was removed, a s shown in Fig. 9.
D ,
FIG.9. The relationship between reaction constant per unit surface bare of oxygen,
and the fraction of surface u free from oxygen. Reproduced from EIey and Rideal, Proe. Roy. SOC.London A178,429 (1941).
It is clear from Fig. 9 that the clean metal surface is an active catalyst for the conversion, and it therefore follows that the most likely mechanism is an exchange of atoms between molecularly adsorbed hydrogen and the underlying primary layer of chemisorbed atoms. This notion was shown to be correct by examining the interchange reaction between a hydrogen covered surface and a low pressure of gaseous deuterium (8). Dz
+ WH
DW
+ HD
178
D. D. ELEY
A clean tungsten surface of sufficient area (400 cm.2) was secured by evaporation from a wire. Hydrogen was admitted and immediately exhausted, leaving a chemisorbed hydrogen film on the tungsten. Deuterium gas was then admitted a t 0.2-mm. pressure. Both at 293" K. and 193" K. a rapid exchange of atoms was found. At 77" K. the equilibrium favored adsorbed hydrogen, and it was necessary to contact hydrogen gas with the surface deuteride, when a rapid exchange was observed. The atomic exchange reaction proceeded with a similar velocity to the paruhydrogen conversion, thus establishing the identity of the latter as a surface exchange reaction. pH1 WH +HW + OH$
+
Certain details are of great interest. All the chemisorbed hydrogen was shown to be active in the surface exchange (and therefore conversion) reaction a t 293" K. From the initial amount of chemisorbed hydrogen one may calculate a film area of 428 cm.2 By measuring the amount of hydrogen appearing in the gas phase with increasing deuterium pressure the results shown in Fig. 10 were obtained. At higher pressures a constant amount of
I 1.0
0.6
0.6 P bm.1
0.4
0.2
FIQ.10. To show how the amount of interchange between gmous DI and a WH surface (D) is affected by the pressure of DI gas. Reproduced from Eley, Proc. Roy. SOC. London A178,452 (1941).
exchange waa obtained which corresponded to the equilibrium shown above being forced completely over to the right. This amount was 2.18XlO'' molecules, and taking l.2XlOI5sites/cm.2, this gave a film area of 364 Since reaction volumes were calculated and not calibrated, the agreement with the first figure is as good as may be expected. At 77" I<. only 20-30% of the chemisorbed deuterium exchanged in the time of experiment, i.e., 6 minutes. Thus, all the sites were not equally active a t this temperature. Since the difference waB undetectable a t 193"K., we may suppose the difference in activation energy between active
CATALYTIC ACTIVATION OF HYDROGEN
179
and inactive planes is not greater than 500 cal./mole. The different sites may be associated with the two common planes 110 and 100. Poisoning the film cut down the amount of exchangeable hydrogen and the conversion velocity in approximately the same ratio. Thus, we see that at temperatures 193' K. all the chemisorbed hydrogen may be taken as active in causing the conversion (or Ha+D2 reaction) by the exchange mechanism. A. and L. Farkas (72) have examined the conversion and surface exchange reactions on evaporated films of nickel, platinum, and palladium. The two reactions have identical rates on nickel at 293" K., while for platinum the exchange is the slower reaction by a factor 5 and for palladium by a factor 15. It is clear that the conversion on nickel goes through the exchange mechanism. In the present author's opinion this conclusion may also be valid for the other two metals, since it is extremely unlikely that two such different mechanisms would have a velocity within a factor as large, even, as 15. The discrepancy might be a zero point energy effect, and a detailed investigation of these metals is planned. Farkas and Farkas refer to a possible dissociation of hydrogen on top of the stable chemisorbed layer, but there are no good physical grounds at present for this postulate. b. Chemisorbed Films of "Foreign" Gases. The parahydrogen conversion on a clean tungsten wire is quite rapid at 142" K., but the admission of a trace of carbon monoxide, nitrogen, oxygen, or ethylene results in the formation of adsorbed films which effectively poison the reaction a t this temperature. On raising the temperature of the poisoned wire, conversion sets in at the following approximate temperatures, carbon monoxide a t 223" K., nitrogen and oxygen a t 273" K., and ethylene at 473" K. For all gases except oxygen the effect was reversible, a return to the lower temperature showing the wire to be still poisoned. Oxygen, however, was removed in some way on heating in hydrogen at 273" K. It is not clear how far conversion on these films proceeds at gaps in the film, or actually on top of the film. Since carbon monoxide may be expected to yield a chemisorbed film without gaps, the latter hypothesis may be true. If true this will be an important development of catalyst theory. Experiments are planned to decide the point. c. Absolute Rate of the Conversion. In a paper submitted for publication, Eley (73) gives a calculation of the "temperatureindependent" factor of the first order constant k. He assumes that: (a) Conversion occurs through an exchange of atoms between a molecule in a van der Waals' layer and an adjacent atom in the underlying chemisorbed layer. (b) The surface atoms are of uniform activity. (c) The van der Wads' layer is a mobile monolayer of hydrogen molecules on localized sites, neighbor interactions being neglected.
>
D. D. ELEY
180
(d) The ratio of partition functions of the chemisorbed atom and the transition complex (Hi) is unity. The method waa to calculate the concentration of molecules in the van der W d s ’ layer by statistical mechanics, and to formulate the rate of exchange by activated complex theory (52). The results for the “temperature-independent” factor B,in k = B . e-E’RT, for a catalyst area of 1cmSg, a reaction volume of 300 cc., and a hydrogen pressure of 20 mm. mercury, are shown in Table 11. TABLE I1 Rate of the Parahydrogen Conversion (ado.)
B mh-1 (exPt.1
170-40 90 100 100
27 110 63 31
Be min. -1 Cat.lyst
w wire
Pt wire Pd disc Ni tube
Much lower values of 13 are found for pyrex glase, which are reasonably explained on the idea of a few sparsely scattered centers of activity. Much higher values of B may be shown to arise for a surface of sites of graded activation energy. Such high values of B are associated with anomalously high values of the activation energy, a case in point being the earliest result of Bonhoaffer and Farkas (74) for a plabinum wire. The detailed mechanism of the interaction between molecular and atomic hydrogen is still an open question. E. K. Rideal (1) haa suggested an elegant mechanism, which makes use of an assumed 8% empty sites in the chemisorbed layer. A hydrogen molecule held over a gap is supposed to undergo a valency switch, the new W-H formed helping to overcome reaction inertia and give a low activation energy. I
H \
H H
D
D
/
D
H
‘ 3 3
W
This mechanism would lead to a mobility of gaps, which, as two gaps came together, would be med up by adsorbing a molecule and thus the reaction would self-poison. On tungsten, however, we might expect the ratio, conversion rate to poisoning rate, to be of the order of 10,OOOand only detectable in specially planned experiments (73). A poisoning action of hydrogen has been described for iron (75) and charcoal (76), but it is not clear how far the conversions observed involved the paramagnetic mechan-
CATALYTIC ACTIVATION O F HYDROQEN
181
ism. The mechanism may be tested by filling the gaps with carbon monoxide or hydrogen atoms, which, on this view, should poison the conversion. The general position regarding the reactivity of a chemisorbed hydrogen atom is quite striking. These atoms, in spite of their strong bond to the tungsten surface, are as reactive a t 77" K. as free gaseous atoms at 900" I<. The present author (73) holds the tentative view that the energy of t,he activated complex is depressed by the action of resonance, a consideration advanced in general terms by Polanyi (77).
VII. THEEXCHANGE OF ATOMSBETWEEN DEUTERIUM GASAND HYDRIDES J. Horiuti and M. Polanyi (78) discovered that platinum black catalyzed the exchange of atoms between deuterium gas and liquid water. A. and L. Farkas and E. K. Rideal (79) discovered the exchange between deuterium and ethylene catalyzed by nickel, to be discussed in section VIII. One of the clearest cases is the exchange reactions of methane and ethane on active nickel catalysts, investigated by Morikawa, Benedict, and Taylor (80). Methane exchanges with methane-dr, deuterium, and deuterium oxide, the rates at 184" C. decreasing in the order given. I t Beems clear that the methane is chemisorbed by rupture of the C-11 bond. The activation energy of 19 kcal. for the exchange with methane-d, is attributed to the energy of desorption of methane. The slower reaction with deuterium has an activation energy of 28 kcal., and the slower rate is attributed to the deuterium being strongly adsorbed and displacing methane from the surface. While ethane exchanges with deuterium at an appreciable rate at 110' C., bond scission producing heavy methane only becomes appreciable a t 150" C. CZH,
+ Da 3 PCHID
There is thus a differenceof 50" C. between the activation of the C-H and the C-C bond, and it is interesting that the weaker C-C bond requires the higher temperature. Farkas and Farltas (81, 82) assumed 'that the rate of dissociation of hydrogen molecules on the catalyst was measured by the parahydrogen conversion. They further assumed that the hydride was adsorbed by dissociation of a hydrogen atom, and that hydrogen and hydride competed for the chemisorbed layer. Hs*H* + H * DZ=D* + D *
I:"d,)
RH *R* + H * RD e R * + D *
The asterisk is meant to indicate a chemisorbed atom or radical. Where
182
D. D. ELEY
the observed conversion was faster than the exchange, the rate of dissociation of hydride (?a) was supposed to be the determining step. Where the two rates were the same, the dissociation of deuterium (lb) limited the rate of exchange. I n the case where the hydride was water, adsorption isotherms were worked out (83). A very good example of their method is given in the case of ammonia (84). The new knowledge about the parahydrogen conversion indicatcs the need for a reconsideration of these views. Much more work is required, but it may be supposed that in the presence of other gases the conversion is proportional to (a) the concentration of chemisorbed hydrogen atoms and (b) the concentration of hydrogen molecules adsorbed in the van der Waals’ layer, or perhaps more specifically on gaps in the chemisorbed layer. The “foreign” gas molecules will compete with hydrogen for sites in the van der Waals’ layer. Competition for the sites in the chemisorbed layer is rather lest3 likely, requiring, as it would, a rather close concordance of two large free energies of chemisorption. At present one might suppose that the chemisorbed layer will consist largely of either the hydride OT hydrogen. Of course, if it is hydride the layer nill contain 50% hydrogen atoms, assuming the R-H bond is broken in the process of chemisorption. The parahydrogen conversion on this view is then, broadly, a test for the presence of chemisorbed hydrogen atoms. The effect of an increased pressure of hydride in decreasing the velocity of parahydrogen conversion is to be attributed in the first place to a displacement of hydrogen by hydride from the van der Waals’ layer. The author does not wish to press these views, but merely to indicate the lines along which his own work will be directed. If we consider the exchange with ammonia (84) it is his view that the chemisorbed layer is likely to be largely either ammonia (held perhaps as H and NH2 fragments) or hydrogen atoms. It is not possibIe to predict which of these two it will be. If wc hold the hydride at constant pressure, and increase the hydrogen pressure a hundredfold, we can only expect a noticeable displacement of chemisorbed hydride if the free energies of chemisorption of hydride arid hydrogen lie within 5 kcal. or so. If this he SO, then a displacement process can occur, perhaps of the kind 112
NII,
I
hI
* 3
M
NHs
‘
hI
H
I
hI
In the case of ammonia exchanging with deuterium on an iron catalyst the facts (84) are: (a) The absolute rate of exchange is independent of the ammonis,
CATALYTIC ACTIVATION OF HYDROGEN
188
pressure and approximately proportional to the square root of the deuterium pressure. (b) Small amounts of hydrogen (but not nitrogen) are formed when ammonia is contacted with an iron catalyst. (c) The conversion of parahydrogen goes rapidly on the clean catalyst. It is reversibly inhibited a t 100" C. by the presence of ammonia, but even then is much faster than the observed exchange reaction. At room temperature it is irreversibly inhibited b y adsorption of ammonia. (d) The activation energy of the conversion is 8 kcal. and the exchange is 15 kcal. Farkas supposcs that mechanism (2b) above, namely, D* + NH,* + NIIzD determines the rate of eschangc, the ammonia being strongly and the hydrogen weakly chemisorbed. This scheme is easily modified to take account of the new view of the conversion and the stability of chemisorbed layers. The ammonia is adsorbed as NH, and H fragments as before. Parahydrogen conversion can proceed, though a t a reduced rate, on such a film by exchange with the chemisorbed hydrogen from the ammonia. KITS pH2
4
NH2*
+ H*
+ H* +11* + ,1111
The exchange reaction may involve some process such a s the following one, making use perhaps of empty sites NH?*
+ Dz
+ NHZD
+ D*
The effect of pressure on the exchange rate wmld then depend on the concentration of deuterium in the van der Waals' layer or over gaps in the chemisorbed layer. Such a view can only be tentative in the absence of further experimental work. Roberts (4) has shown how oxygen will displace chemisorbed hydrogen from tungsten. Here we pass from a full hydrogen layer to a full oxygen layer, the latter being more firmly bound. We require more work along this line for other gases, before we can bc sure of our catalytic mechanisms. For the same reasons, the exact mechanism of the exchange reaction with liquid and gaseous water (7, 81, 83, 85, 86) is still an open problem. The notion that a saturated hydride KH, when it is chemisorbed, is held as R and H fragments seems fundamentally sound. So far as the present author is aware, an appreciable parahydrogen conversion is always obtained in such cases, showing the presence of chemisorbed hydrogen. This is in direct contrast to chemisorbed ethylene and carbon monoxide (see
D. D. ELEY
184
below) , where the parahydrogen conversion may be almost completely inhibited.
VIII. EXCHANGE AND HYDROQINATION OF UNSATURATED HYDROCARBONS The hydrogenation of ethylene by deuterium was shown to be accompanied by an atomic exchange (79), and a similar exchange reaction was found with benzene (87). At present three viewpoints may be distinguished (1) The original viewpoint of Horiuti and Polanyi (88) was thqt ethylene wa8 adsorbed by the opening of the double bond. Attack by one chemisorbed deuterium atom resulted in a half-hydrogenated state. This could revert to chemisorbed ethylene, losing a hydrogen atom to give heavy ethylene, or attack by a second deuterium atom could result in heavy ethane. On this view exchange and hydrogenation reactions are intimately connected.
+
*CH&H,* D* -+*CH&HzD (Hall-hydrogenated state) *CHzCHID+ *CHzCHD* H* (Exchange) *CH&H,D -+ C&DCILD (Hydrogenation)
+
The asterisk at each end of the ethylene is meant to indicate the two adsorption bonds, one from each carbon, to two surface metal atoms. (2) Twigg and Rideal (89) accept the “half-hydrogenated state” a8 sn intcrmediate in the exchange reaction, only modifying Polanyi’s picture in certain details. Hydrogenation is viewed as a separate reaction. (3) A. Farkaa (82) maintains that ethylene, like the saturated hydrides, is adsorbed by rupture of a C-H bond, exchange occurring in the manner mentioned in section VII. CHFCII~ --+ CIIi=CH* + H* Dn+D* + D * D* CI€2=CEI* +CH,=CHD.
+
Exchange mechanisms (1) and (2) have been called “associative,” and mechanism (3) has been called “dissociative.” In their original paper Polanyi and Horiuti (88) noted that the occurrence of cis-trans conversion and double bond migration definitely favored their “dissociative” mechanism for ethylene adsorption. The extensive work of Twigg and Rideal which will now be described has borne out their conclusion. In the first place (89) they made a careful study of the kinetics of the exchange between deuterium and ethylene on a nickel catalyst. Taylor and coworkers had shown that on a specially active nickel catalyst this exchange was detectable at -8OOC. (go), while the exchange with ethane only set in at 100” C. (80). Twigg and Rideal found that in the presence of ethylene the H,+Da reaction was completely inhibited at temperatures up to 160”C., pointing to an absence of chemisorbed deu-
185
CATALYTIC ACTIVATION O F HYDROGEN
terium atoms. At 156' C. the exchange and hydrogenation rates were both independent of ethylene pressure and directly proportional to deuterium pressure. These results thus point to a very strong chemisorption of ethylene by opening of the double bond. Since the ethylene is assumed to be adsorbed at two points, it follows that there will be gaps in the monolayer, and they suppose that the exchange involves deuterium molecules adsorbed on these gaps. CH2-CIIs
Da
Ni
Ni-tki
l
l
Ni
CH2-CH2D
=
Ni
D
CH2-CHD
I
hii+Ni
1
Ni
HD Ni
Or in our shorthand notation *CHKHz*
1 + Dz + *CHaCH*D + D*
2 --j
*CHzCHD*
+ HD
The apparent activation energy of 18.6 kcal. was attributed to step 1 or 2, the adsorption of desorption of deuterium. A chemisorbed atom, D*, is believed to exchange very rapidly with hydrogen atoms in neighboring ethylene molecules. According to this view a deuterium molecule following the initial act of adsorption will come to equilibrium with six neighboring CH2groups before desorption and thus the hydrogen desorbing from the catalyst in the initial stages of reaction should contain 14% deuterium. They found 20% deuterium, in reasonable agreement with calculation. Conn and Twigg (91) found that no exchange occurred between ethylene and ethylene-ds on a nickel catalyst. Twvigg (92) further showed that double bond migration in the case butene I+ butene I1 occurred on his nickel catalyst, but only in the presence of hydrogen. *CHa*CHCHzCHa
+ D*
+
CHaDC*HCHzCH, $ CHzDC*HC*HCHs H*
On the dissociative mechanism it would reasonably be expected that only hydrogen atoms which are dissociated at the relatively low temperature of experiment, i.e., those adjacent to the double bond, would exchange with deuterium. Twigg (93) found, however, that all the hydrogen atoms in propylene and butylene were exchangeable, aa may be understood by consideration of the associative mechanism for double bond migration above. Using data on bond lengths and bond angles, Twigg and Rideal (94) calculated that a spacingof 2.73A. would be justWrighttdgive two point biradical. 'The active catalysts have spacadsorption of the -CH&H2-ings in the neighborhood of this value. They showed that while ethylene molecules could pack so as to fill the nickel surface and exclude chemisorbed hydrogen, the bulky nature of the methyl ethylenes would necessarily result in bare sites which could chemisorb hydrogen. In agreement with their calculation, they found that an H2+D2 reaction took place in the presence of the substituted ethylenes.
186
D. D. ELEY
This extensive work finds a natural explanation in terms of the associative mechanism, though A. Farkas has offered interpretations of some of the results in terms of his dissociative mechanism. He suggested (95) that thc absence of exchange in ethylene-ethylene-d4 was due to a low concentration of chemjsorbed hydrogen. He has also proposed a mechanism for double bond migration (82). His remarks, particularly on double-bond migration, have been criticized by Twigg (96), who has also pointed out the necessary critical experiment. If ethylene is adsorbed in the way Farkas suggeated, hydrogen should desorb and be detectable in the gas phase. This experiment has not yet becn made. There are some grounds for regarding hydrogenation as a separate problem, and its mechanism is still obscure. A. Farkas (82) has shown that in the catalytic hydrogenation of acetylenic or ethylenic bonds the products formed arc not necessarily those of the highest thermodynamical stability, but those which would result from cis-addition of the hydrogen. He concluded that hydrogen atoms are not added independently, but that both thc hydrogen atoms are added more or less simultaneously from one molecule. This argument would not seem to be valid, and Greenhalgh and Polanyi (97) have shown how the independent addition of hydrogen atoms to the -CH&H2- biradical, via the half-hydrogenated state, would also give cisaddition. Twigg and Rideal (89) found that the activation energy for hydrogenation of ethylene was always 4-5 kcal. less than that for the exchange reaction. Since exchange is limited by the adsorption or desorption process, e.g., *CIIzCHz*
+ D1+
*CH&H1D
+ D*
they supposed the hydrogenation rate to be limited by the addition of molecular hydrogen across the double bond. They showed that the wellknown decrease in the temperature coefficient for the hydrogenation was paralleled by a similar effect for the exchange reaction. The decrease sets in a t 100' C., which is the temperature nt which the desorption of hydrogen commences. Greenhalgh, Horrex, and Polanyi (97, 98) maintained Polanyi's original view that the half-hydrogenated state, *CH&H,, is a half-way house to ethane, reached by the independent attachment of a second chemisorbed hydrogen atom. On their view they consider that the apparent order of the exchange reaction should be one-half that for the hydrogenation, and find this is so for a number of experiments with benzene. Their papers should be consulted for the detailed treatment of a large number of results on the parahydrogen conversion, exchange, and hydrogenation. They predicted that the partial hydrogenation of methyl oleate with deuterium should result in isomcrs with shifted double bonds, all such isomcrs containing an atom of deuterium. Baxendale and Warhurst (99a) actually
CATALYTIC ACTIVATION O F HYDROGEN
187
found the main product of this experiment t o be light elaidic acid. While obtaining no positive evidence, they considered that their result is at least reconcilable with the associative mechanism, and is definitely against the dissociative mechanism. Balandin (99b) in a recent paper assumes that hydrogen and the substrate are adsorbed on two different kinds of site. Exchange and hydrogenation are limited in rate by the formation of the half-hydrogenated state, through the migration of a hydrogen atom between the two kinds of center. He discusses in detail published kinetic data from this point of view. 0. Beeck (33d) gives a short account of his extensive investigations on the hydrogenation of ethylene on evaporated films of metal, particularly nickel. We can only make tentative comments on this work, since full details have still to appear, and a monograph is in preparation by Beeck and Wheeler. Beeck's values for the heats of chemisorption for hydrogen on nickel are 30-18 kcal., and for ethylene 60-25 kcal., the heats decreasing with surface covering. These figures would support the Rvigg-Rideal view that ? nickel surface exposed to a mixture of ethylene and hydrogen is largely covered by chemisorbed ethylene, and indeed it would seem quite possible to reconcile Beeck's kinetic data on the hydrogenation with this view. However, Reeck takes quite a different line. He has obtained evidence that ethylene is largely adsorbed as acetylenic complexes. These block the surface and in his view limit the rate of reaction, which is determined by the amount of hydrogen chemisorbed on the fraction not covered by acetylenic complexes. Experimentally he finds that ethylene will pick off hydrogen atoms from a preadsorbed film very rapidly, and he thus considers that reaction occurs by collision of ethylene from the gas phase with the chemisorbed hydrogen. At first sight, this view wouId conflict with Twigg and Rideal's observation on the inhibition of the H2+D2 reaction, but further discussion a t present would be premature. Besides these conclusions on the kinetics of the reaction, which are undoubtedly controversial, Beeck presents interesting data on the relative efficiencies of the various transition metals. Rhodium turns out to be the best catalyst, which he associates with its 3.75 A. spacing. IX. THEFISCHER-TROPSCH SYNTHESIS
This provides an interesting example of the diagnostic value of the parahydrogen conversion. In the normal synthesis carbon monoxide and hydrogen are passed over a promoted iron or cobalt catalyst at atmospheric pressure and 200-250" C., long chain hydrocarbons being formed. Under these conditions Craxford (100) found the parahydrogen conversion to be
188
D. D. ELEY
inhibitcd. Thus there is little chemisorbed hydrogen on the catalyst and Craxford gave reasons. for supposing, in the first instance, that a cobalt carbide surface is formed by the reaction 2C0 + C X I + H,+
X. DEHYDROGENATION AE was pointed out earlier, the mechanism of hydrogenation is still obscure, and since there has been little work in recent years on dehydrogenation, we can only discuss possible mechanisms in the most general way The reaction is normally conducted a t sufficiently high temperatures t o assume chemisorption of the substrate, e.g., ethane, by rupture of the C-H bond. We would imagine two successive steps CHsCHs +CHsCH1*
+ €I* +*CH&HZ* + 2H*
At temperatures of 400' C. or so, the desorption of ethylene and hydrogen will proceed fairly rapidly on most catalysts. There would seem no reason to postulate simultaneous release of two hydrogen atoms to form a' molecule, i.e., the reverse of the hydrogenation mechanism, since conditions in the chemisorbed layer will be different from those Twigg and Rideal found for the hydrogenation a t 100" C. Since chemisorbed hydrogen is expected, bond migration will probably occur simultaneously. Certain classical work is in accordance with this view. Thus Palmer and Constable (102) showed that the five alcohols, ethyl, propyl, butyl, isobutyl, and isoamyl, had the same rates of dehydrogenation on a copper catalyst under the same conditions. The energy of activation was about 22 kcal. They supposed that the alcohols were adsorbed with their chains vertical to the catalyst surface, and that the rate determining step was the removal of a hydrogen atom from the hydroxyl group (103). The rearrange-
CATALYTIC ACTIVATION OF HYDROGEN
189
ment of bonds then expelled the second hydrogen atom from the carbon atom without activation energy. We would, however, imagine that the second step would be adsorption at the carbon atom with rupture of a carbon-hydrogen bond, though the activation energy for this second step might be small. Balandin (104) has considered that two point contact is necessary in the dehydrogenation of ethyl alcohol, this condition defining the necessary lattice spacing in the catalyst (105). His view again is in agreement with the above pioture, if we suppose that the contact at two points does not need to occur simultaneously. Schwab (106) has examined the dehydrogenation of formic acid on metal alloy catalysts, paying particular attention to homogeneous solutions (a phases) of silver with other metals. Except where the second metal was itself an active catalyst (platinum, palladium, gold), the added metal increased the activation energy of the dehydrogenation process, the increase being greater the greater the valency of the added metal. More specifically,the increase in activation energy was related to Hume-Rothery's electron concentration function for the solute metal, i.e., the number of valency electrons divided by the number of atoms. Thus, as the Brillouin zone of the alloy fills up with electrons, the activation energy increases. Additional support came from considering the complete series of phases formed as one metal, for example antimony, was added to silver. According to H. Jones, as the Brillouin zone becomes filled with electrons the phase becomes unstable. At the phase change a new lattice is formed, with a more capacious zone to accommodate further electrons. Schwab shows that the phase change is accompanied by a fall in activation energy for the alloy as a catalyst. His conclusion is that the primary step in the catalysis is the loss of two electrons to the Brillouin zone of the metal, the two hydrogen ions left subsequently slipping into interlattice places near the metal surface.
XI. CYCLIZATION OF WPARAFFINS In the presence of oxide catalysts such as chromium oxide at 5OO0C., n-heptane will cyclize to toluene. There are also many other possible reactions, and Taylor and Turkevich (107) have examined the thermodynamics of the reaction and its alternatives. They concluded that cyclization was the favored reaction above 300' C., and pointed out the need to choose a catalyst which activated the C-H rather than the C-C bond. Pitkethly and Steiner (108) reacted n-heptane on a dehydrogenating catalyst in the presence of hydrogen at 475" C. For a constant contact time they found a constant heptene to heptane ratio in the products, when the partial pressure of hydrogen was varied from 120 to 495 mm. mercury.
190
D. D. ELET
Increased contact time sent up the yield of toluene, but left the heptene yield unchanged. These results suggested to them that heptene was an intermediate in the production of toluene and mas present in a stationary concentration.
The methyl cyclohexene is subsequently dehydrogenated to toluene. Herington and Rideal (109) consider that the intermediate heptene is held by two point contact at the double bond, as in the above formulas. The double bond may migrate to a favorable position for ring closure t o occur. They investigated molybdenum oxide as catalyst. They associated the catalytic activity with a 2.79 A. spacing of the molybdenum atoms not present in the metal itself, and this enabled them to make a simple statistical calculation of the variation of total yield of aromatics with the nature of the paraffin feed, in agreement with experiment. Their failure to calculate the distribution of the different aromatics they attributed to isomerization on the catalyst. m-Xylene and p-xylene, for example, were found in the products from n-octane, and these could only arise by an isomerization accompanying the actual ring closure. They supposed this to involve the formation and rupture of a subsidiary ring, as below.
XII. THEHTDROGEN-OXYGEN REACTION While this reaction was one of the first to be investigated, its mechanism is still obscure. Benton and Elgin (110) found for a silver catalyst that the reaction rate mas independent of the oxygen pressure and directly proportional to the hydrogen pressure. These kinetics fitted in with their adsorption measurements, which showed zero adsorption of hydrogen but an appreciable adsorption of oxygen. They supposed that gaseous hydrogen molecules attacked the adsorbed oxygen, and to account for the slow rate observed, suggested that only oxygen atonis next to bare spaces could react.
CATALYTIC ACTIVATION O F HYDROGEN
191
Such a picture would fall in with the general viewpoint taken in this paper. However, while gold (111) showed a similar adsorption behavior, the reaction rate was proportional to the square of the hydrogen pressure and directly to the oxygen pressure. Again, investigations of nickel (112), silver (113) and platinum (114) in Chapman’s laboratory have emphasized that the chemisorbed film is mainly oxygen, and the view has been expressed that the reaction goes through an alternate oxidation and reduction of the surface. The complex nature of these reactions may, in part, reside in the tendency for certain metals t o give surface oxide layers several molecules thick which may, or may not, be catalysts for the reaction. Thus, while silver or gold is poisoned by a thick oxide layer (115), this does not seem to be the case for nickel (112), which, in the presence of a hydrogenaxygen mixture, is reported to be covered by an oxide film 20A. thick. Bulk nickel oxide is known to be a catalyst (116).
XIII. BIOLOGICAL ACTIVATION OF HYDROGEN The biological activation of hydrogen in organic compounds, forming, as it does, the first step in biological oxidations, is a vast field, which we cannot attempt t o review here. The few remarks that follow are only intended to relate this field to the activation of hydrogen by metals, as it is quite likely that there is a common basis for catalytic activity in the two cases. H. Wieland (117) showed that dehydrogenation mas the first stage in the enzymatic oxidation of organic molecules. Thus, paIladium black catalyzed the oxidation of ethyl alcohol not only by oxygen but also anaerobically by quinone or methylene blue. The palladium black could be replaced b y dehydrogenases, and the first step is clearly RHzfEnzyme+R f Enzyme-Ha
It is now known that the enzyme may be separated b y dialysis into a coenzyme and a protein. The protein is the true catalyst and is specific for each substrate. Its function is to combine Kith both substrate and coenzyme, when it can mobilize the hydrogen for transfer from the one t o the other. The combinations are supposedly reversible, and have been described by two equilibrium constants D and d (118). Protein +Cocnl,yme,Protrin-Cocnzyme, D
+
Protein-Coenzyme SirbstrateeProtein-Coenzyme-Substrate, d
A first order velocity constant, k, served to describe the actual hydrogen transfer within the complex. Two coenzymes have been isolated and their structures determined.
192
D. D. ELEY
Coenzyme I (cozymase) acts with alcohol and some other substrates, while coenzyme I1 acts with glucose 6 phosphate. They are both phosphopyridine nucleotides (119), the active group in hydrogen transfer being the pyridinium ring (120).
I
d
R
D , d , and k have been determined by Warburg from an analysis of the kinetics of the reaction. The hydrogen is next transferred to the cytochrome reductase molecule, the prosthetic group of which is alloxrtaine (121). In this case there is definite evidence that the two hydrogen atoms are added separatcly, the alloxazine passing through a semiquinone stage (122). R
R
I
F
I
N
N
++H c C IE HI aa \/Aw/ d\Nc
N II
CHI CH, \ o(J)C=O
c: II
N
I
0
OH
N
/
+n -+
C
I
OH
R
I
N
NH
=O
Michaelis and Schubert (123) have represented dehydrogenation as a removal, independently, of electron plus hydrogen ion, rather than hydrogen atom. While this concept is no doubt physically exact for ionizable compounds like hydroquinone, it would seem to have a purely formal use for substrates such as succinic acid, where the hydrogen atoms concerned do not ionize. The concept, however, serves a useful purpose in unifying the dehydrogenase action with the oxygen end of the enzyme chain, where the redox changcs are simple electron transfers between the iron containing cytochrome molecules. Two enzymes are found in certain bacteria, such as Escherichia coli, which will activate gaseous molecular hydrogen (124). Both have been shown to catalyze forward and back reactions. Hydrogenase catalyrea H,+Ai=AH2, where A is an acceptor such aa methyl viologen, or even a
CATALYTIC ACTIVATION OF HYDROGEN
193
one-carbon compound. Hydrogenlyase catalyzes the product ion of formic acid, Ha+C02eHCOOH. E . coli has also been shown t o catalyze the exchange of atoms between deuterium and water (125, 126). Huberman and Rittenberg (126a) find that Proteus vulgaris contains an enzyme that catalyzes the exchange reaction H2 Da. The enzyme is inhibited by CO and HCN and they tentatively conclude that it is a ferrous heme system.
+
XIV. GENERALCONCLUSIONS We are scarcely at a stage where any generally applicable theories of either the kinetics or the energetics of catalysis may be applied with any real confidence. As the present author sees it, the need is still for thorough studies of specific systems, using combinations of chemical and physical techniques. Little advantage has been taken of the methods which Langmuir made available 30 years ago for the detailed study of catalytic reactions under clearly defined conditions, and the tendency has been to work with complex surfaces, where “active spots’, may be expected to playan important role. This notion of “active spots” must remain a relatively sterile hypothesis until the characteristics of uniform surfaces have been more fully explored. With further data of this kind at our disposal we could more confidently approach the problems presented by the physical (127) or chemical (128) heterogeneities of activated “industrial” catalysts. So the remarks in this section may only be regarded as the personal orientation of the author to what is still a predominantly experimental field of research. 1 . Stability of the Chemisorbed FiZm It is now clear from Roberts’ work that chemisorbed films of hydrogen and other gases may be much more stable than was generally supposed. They may still, however, be very reactive, so that suggestions (129) that catalytically active hydrogen is weakly chemisorbed cannot be generally applicable. It seems clear also that the Langmuir-Hinshelwood picture (3) of catalysis in a simple competitively occupied chemisorbed layer must be modified in such cases. The Rideal picture (2) of interaction between a van der Waals layer and a chemisorbed layer has so far been applied with some success to a number of reactions, and probably has a fairly wide application to reactions at relatively low temperatures. The domain of the Langmuir-Hinshelwood picture would seem to be reactions at relatively high temperatures, where desorption of the chemisorbed layer has set in.
2. Geometry of the Catalyst Surface Roberts has shown on theoretical grounds how gaps may arise in immobile and mobile monolayers. While much remains to be done in the experi-
194
D. D. ELEY
mental investigation of these gaps, the concept would seem to be of the greatest importance for catalysis. Herington and Rideal (130)have shown how the action of bulky poisoning molecules in excluding the occupation of neighbor sites in a uniform lattice can give rise tononlinear poisoning curves which have previously bcen explained on the active spot theory. Some success has also been achieved with the notion that the spacings in a lattice active in the catalysis of ethylene and acetylene are those which can chemisorb the biradical. In this connection the experimental work of Beeck and coworkers (9) is of the greatest importance. They showed how oriented metallic films could be prepared by evaporation from wires in the presence of inert gas and demonstrated in this way that the 110-plane of nickel was more active in the hydrogenation of ethylene than the other planes. A great deal more work along this line is required before we can get a clear picture. The work would suggest that even an aged wire might exhibit more than one zone of activity, depending on the occurrence of more than one lattice plane in the surface. 3. Energetics of the Transition State
A catdysed reaction follows a path of relativeIy Iow free energy of activation, in the sense of Eyring's theory, as compared with the same reaction proceeding without a catalyst. Where a comparison is possible, it would seem that inorganic catalysts lower the energy (heat content change) of activation (3), and the same holds for enzymes (131). The effect of catalysts on the entropy of activation, so far as the present author ie aware, is, in general, less marked. I n some way the molecules adsorbed on the metal, metal oxide, or other catalysts, or held to a specific protein, are converted into a more labile form, i.e., the potential hill for one specific mode of reaction (out of perhaps a number of such) is considerably lowered. It has long been knoum that mere adsorption is not sufficient. Indeed, tungsten adsorbs ethylene very strongly and possesses just the right 2.73A. spacing to hold it in the biradical form, yet it does not seem to be an efficient catalyst for the exchange reaction between hydrogen and ethylene (28). In the enzyme field Quastel (132) has shown how a dehydrogenase is inhibited if molecules of related structure to the specific substrate are added to the system. These molecules are, indeed, adsorbed by the enzyme, or perhaps more correctly, form the enzyme-substrate complex, because this is largely a stereochemical matter. They are not, however, brought into reaction. Thus various acids containing the CHZCOOH group, such as malonic and glutaric acids, will inhibit the dehydrogenation of succinic acid.
195
CATALYTIC ACTIVATION OF HYDROGEN
It is not clear how far all catalysts, metals, metal oxides, and proteins may possess a common bMis for their activity, but it seems reasonable to take this view as a working hypothesis for the present. Stearn (131), for example, has supposed that the charged groups present in proteins lower potential barriers for adsorbed substances by increasing the ratio of coulombic to exchange energy in the bonds concerned. Such an effect follows from the London-Eyring-Polanyi treatment of potential energy barriers. Again, we have referred to Polanyi's notion (77) concerning the effect of resonance energy in depressing the activation energy for a reaction. This idea has so far only been developed for the diene synthesis (133) , and it can scarcely be applied to catalysis until we possess a more detailed picture of at least one reaction path, which is a matter for experiment. Theoretically, too, we need to know more about the energy levels and wave functions of the surface electronic states. All three types of catalyst possess common energy levels, i.e., the valency electrons of the atoms are not localized but may move through the lattice (excitation energy being necessary for nonmetals). Metals are electronic conductors, and certain of the catalytic oxides, e.g., nickel oxide, are known semiconductors. Szent Gyorgyi (134)has recently shown that gelatin impregnated with a dye is a photoconductor. It is not, indeed, an easy matter to formulate the problem in any but a vague way by the electron pair method of Pauling. The whole problem of transition states adsorbed on metals is probably better considered in terms of the molecular orbital method of LennardJones and Coulson (135) and others, since this method is closely related to the method used for metals. A virtue of the gap hypothesis for catalysis is that we can visualize the transition state as a ring of mobile electrons. H* M*
H*
H* M*
Parahydrogqn conversion
*CHZ
&
CHZ'
H*
&
H*
&
Ethylene-hydrogen exchange
M indicates a surface atom of the metal and the asterisk a mobile electron. The resonance energy might be expected to strongly depress the energy of formation of such a complex from the initial adsorbed reactants, i.e., the activation energy for the reaction. Considerations of this kind, however, still fall into the realm of speculation. REFERENCEB 1. Rideal, E. K., Proc. Cambridge Phil. Soc. 36, 130 (1939). 2. Rideal, E.K., Sabatier Lecture, Chemistry & Industrv 62,335 (1943). 3. Hinshelwood, C. N., Kinetics of Chemical Change. Oxford University Press, London, 1940.
196
D. D. ELEY
4. Roberta, J. K., Proc. Roy. SOC.London A152,445 (1935). 5. Roberts, J. K., Some Problems in Adsorption. Cambridge University Press, London, 1939. 6. Bosworth, R. C. L., and Rideal, E. K., Proc. Roy. SOC.London A162, 1 (1937). 7. Bosworth, R. C. L., Proc. Cambridge Phil. Soc. 33,394 (1937). 8. Eley, D. D., Proc. Roy. SOC.LondonA178, 452 (1941). 9. Beeck, O., Smith, A. E., and Wheeler, A,, Proc. Roy. SOC.London A177, 62 (1940). 10. Van Cleave, A. B., Trans. Faraduy Soc. 34,1174 (1938). 11. Morrison, J. L., and Roberts, J. K., Proc. Roy. SOC.London A173, 1, 13 (1939). 12. Roberta, J. K., Nature 137, 659 (1936). 13. de Boer, J. H., Electron Emission and Adsorption Phenomena. Cambridge University Press, London, 1935. 14. Surhmann, R., and Csech, H., 2.physik. Chem. B28, 216 (1935). 15. Leypuneky, 0. I., A& Physicochim. U.R.S.S.2,737 (1935). 16. Langmuir, I., and Kingdon, K. H., Phys. Rcu. 36, 129 (1929). 17. Bosworth, R. C. L., and Rideal, E. K., Physica 4,925 (1937). 18. Rupp, E., Z . Elektrochem. 36, 686 (1929). 19. Germer, L. H., 2.Phyeik 64,408 (1929). 20. Wolfenden, J. H., Proc. Roy. Soc. London A110,464 (1925). 21. Kistiakowsky, 0. B., J. Phys. Chem. 30, 1356 (1926). 22. Taylor, H. S., and Kistiakowsky, G. B., 2.physik. Chem. 126,341 (1927). 23. a. Eley, D. D., Nature, 168, 449 (1946). b. Frankenburger, W., and Mayrhofer, K . , Z . Eleklrochem. 36,590 (1929); Frankenburger, W.,Mayrhofer, K., and Schwamberger, E., ibid. 37, 473. (1931). 0 . Langmuir,I., J . Am. Chem. SOC.36,931 (1913);Z. anorg. Chem. 83,261 (1914). d. Fraukenburg, W. G., J . Am. Chem. SOC. 66,1827,1838 (1944). 24. Bosworth, R. C. L., Trans. Faraday SOC.36,397 (1939). 26. Taylor, H. S., J . Am. Chem. SOC.63, 678 (1931); Chem. Reu. 9, 1 (1931); Trans. Faraday SOC.28,131 (1932). 26. Frankenburger, W., and Hodler, A., Trans. Faraday Soc. 28,229 (1932). 27. Davis, R. T., J. Am. Chem. SOC.68, 1396 (1946). 28. Eley, D. D., and Rideal, E. K., Proc. Roy. SOC.London 178,429 (1941). 29. Langmuir, I., Phys. Rev. 22, 374 (1923). 30. Johnson, R. P., Phys. Rev. 64,459 (1938). 31. Pauling, L., The Nature of the Chemical Bond. Cornell University Press, Ithaca, 1939. 32. Roberta, J. K.,Some Problems in Adsorption. Cambridgo University Press, London, 1939, pp. 88, 113. 33. a. Taigg, G. H. and Ridcd, E. K., Trans. Faraday Soe. 36,533,1940. b. Tnylor, J. B., and Langmuir, I., Phys. Rm.44,423 (1933). c. IIalsey, G., and Taylor, H. S., J. Chem. Phys. 16,624 (1947). d. Beeck, O., Rev. Modern Phys. 17, 61 (1945). 34. Lennard-Jones, J. E., Trans. Faraday Soc. 28,333 (1932). 35. Wang, J. S., Proc. Roy. SOC.LondonA161,127 (1937). 36. Miller, A. El., and Roberte, J. K., Proc. Carnbrh’ge Phil. Soc. 37,82 (1941). 37. Eyring, H., and Sherman, A., J. Am. Chem. Soc. 64,2661 (1932). 38. Okamoto, G., Horiuti, J., and Hirota, K,, Sci. Papers Znst. Phys. Chem. Research (Tokyo) 28,223 (1936). 39. Eyring, H., and Polanyi, hf., 2. physik. Chem. Bl2, 279 (1931). 40. Lanamuir, I., Chem. Revs. 13, 147 (1933).
CATALYTIC ACTIVATION OF HYDROGEN
197
41. de Boer, J. H., Electron Emission and Adsorption Phenomena. Cambridge University Press, London, 1935,p. 76. 42. T a m , I., Physik. Z. Sowjetunion 1, 733 (1932). 13. Shockley, W.,Phys. Rm. 66,317 (1939). 44. Pollard, W.G., Phys. Rev. 66,324 (1939). 45. Polanyi, M., Z.Elektrochem. 36, 561 (1929). 46. Coehn, A., and Jurgens, H., 2. Physik 71, 179 (1931). 47. Franck, J., Nuchr. Ges. Wiss. G Ittingen, 293 (1933)(quoted in ref. 13). 48. Langmuir, I., J . A m . Chem. Soe. 34, 1310 (1912);ibid. 37,417 (1915). 49. Bryce, G.,Proc. Cambridge Phil. SOC.32,648 (1936). 50. Roberta, J. K., and Bryce, G., Proc. Cambridge Phil. Soc. 32,653 (1936). 51. Giauque, W.F., J . Am. Chem. SOC.62,4816 (1930). 52. Glasstone, S., Laidler, K. J., and Eyring, H., The Theory of Rate Processes. McGraw-Hill, New York, 1941. 53. Bonhoeffer, K. F., Z. physik. Chem. 113,199 (1924). 54. Roginsky, S.,and Schechter, A., Acta Physicochim. U.R.S.S.1,318 (1934). 55. Scheehter, A.,Actu Physicwhim. U.R.S.S. 10,379 (1939). 56. Bonhoeffer, K.F.,and Harteck, P., Z . physik. Chem. B4, 122 (1929). 57. Bonhoeffer, K.F.,Farkas,A., and Rummel, K. W., 2. physik. Chem. B21,225 (1933). 58. Farkaa, L.,and Sachsse, H., 2. physik. Chem. B23,1,19 (1933). 59. Wigner, E., Z.physik. Chem. B23, 28 (1933). Sce also Kalekar, F., and Teller, E., Proc. Roy.SOC.London A160, 520 (1935). 60. Taylor, H.S.,and Diamond, H., J. Am. Chem. SOC.67,1251 (1935). 61. Bonhoeffer, K. F., and Farkas, A., T r a m . Faraday SOC.28,242(1932). 62. Taylor, H.S.,and Sherman, A., Trans. Faraday SOC.28,247 (1932). 63. Fajans, E., Z.physak. Chem. B28,239,252(1935). 64. Gould, A. J., Bleakney, W., and Taylor, H. S., J. Chem. Phys. 2, 362 (1934). 65. Farkaa, A., and Farkas, L.%Nature 132,894 (1933). 66. Farkaa, A., Orthohydrogen, Parahydrogen and Heavy Hydrogen. Cambridge University Prws, London, 1935. 67. Burstein, R., Aciu Physicochim. U.R.S.S.8,857 (1938). 68. Turkevich, J., and Selwood, P. W., J . Am. Chem. SOC.63,1077 (1941). 69. Farkaa, A., Z.physik. Chem. B14, 371 (1931). 70. Roberta, J. K., Tram. Faraday SOC.36,941 (1939). 71. Farkas, A., Trans. Faraday SOC.36,943 (1939). 72. Farkaq A., andFarkaa, L., J. Am. Chem. SOC.64,1594 (1942). 73. Eley, D.D.,Trans. Faraday SOC.44,216 (1948). 74. Bonhoeffer, I(.F., and Farkas, A., Z . physik. Chem. Bl2,231 (1931). 75. Emmet, P. H., and Harknesa, R. W., J. A m . Chem. Soe. 67,1624(1935). 76. Burstein, R.7., and Kaahtanov, P., Trans. Faraday Soc. 32,823 (1936). 77. Polanyi, M., Sci. J. Royal Coll. Sci. 7,21 (1937). 78. Horiuti, J., and Polanyi, M., Nature 132,819 (1933). 79. Farkaa, A., Farkaa, L., and Rideal, E. K., Proc. Roy. SOC.London, A146,630 (1934). 80. Morikama, K., Benedict, W. S., and Taylor, H. S., J . A m . Chem. Sor. 68,1445,1795 (1936). 81. Farkas, A.,and Farkes, L., Trans. Faraday SOC.33,678(1937). 82. Farkaa, A., Trans. F a r d a y Soc. 36,906 (1939). 83. Farkas, A., Trans. Faraday SOC.32, 922 (1936). 84. Farkas, A., Trans. Faraday SOC.32,416 (1936). 85. Horiuti, J., and Polanyi, hf., Mem. Proc. Manchestcr Lit. h Phzl. SOC.78, 47 (1934).
198
D. D. ELEY
86. Eley, D. D., and Polanyi, M., Trans. Faraday SOC.32, 1388 (1936).
87. Horiuti, J., Ogden, G., and Polanyi, M., Trans. Faraday SOC.30, 663 (1934). 88. Horiuti, J., and Polanyi, M., Trans. FaradaV Soc. SO, 1104 (1934). 89. Twigg, G. H. and Rideal, E. K., Proc. Roy. Soc. London, A171,55 (1939). 90. Morikawa, K., Trenner, N. R., and Taylor, H. S., J. Am. Chem. Sac. 69, 1103 (1937). 91. Conn, G. K. T., and Twigg, G. H., Proc. Roy. Soc. London, A171, 70 (1!139). 92. Twigg, G. H., Proc. Roy.SOC.LondonA178,106 (1941). 93. Twigg, G. H., Trans. Furaday SOC.36,934 (1939). 94. Twigg, G. H., and Rideal, E. K., Trans. Faraday SOC.36,533 (1940). 95. Farkaa, A., Trans. Faraday Soc. 36,941 (1939). 96. Twigg, G. H., Tram. Faraday SOC.36,945 (1939). 97. Greenhalgh, R. K., and Polanyi, M., Tram. Faraday Soc. 56,520 (1939). 98. Horrcx, C., Greenhalgh, R. K., and Polanyi, M., Trans. Faraday SOC.36,511 (1939). 99. a. Baxendale, J. €I., and Warhurst, E., Trans. Faraday SOC.36, 1181 (1940). b. Balandin, A. A., ActaPhysic-chim. U.R.S.S. 22,80 (1947). 100. Craxford, S. R., Trans. Faraday SOC.36,948 (1939). 101. Herington, E. F. G., Trans. Faraday Sor. 87,361 (1941). 102. Palmer, W. G., and Constable, F. H., Proc. Roy. Soe. London A107,2.55 (1925). 103. Constable, F. H., Proc. Roy. SOC.London A108,355 (1925). 104. Balandin, A. A., Z . physik. Chem. B2,293 (1929). 105. Balandixi, A. A., and Zelinsky, N. D., Z . physik. Chem. 126, 267 (1927). 106. Schwab, G. M., Trans. Faraday SOC.42, 689 (1946). 107. Taylor, H. S., and Turkcvich, J., Tram. Faraday SOC. 36,921 (1939). 108. Pitkethly, R. C., and Steiner, H. S., Trans.Furaday SOC.36,979 (1939). 109. Herington, E. F. G., and Rideal, E. K., Proc. Roy. SOC.LondonA184,434,447 (1945). 110. Benton, A. F., and Elgin, J. C., J. Am. Chem. SOC.48,3027, (1926). 111. Bentou, A. F., and Elgin, J. C., J . Am. Chem. SOC.49,2426 (1927). 112. Hughes, n.R., and Bevan, D. C., Proc. Roy. Soe. London A117, 100 (1927). 113. Hall, W. K., and Chapman, D. L., Proc. Roy. Sac. LondonAl24,478 (1929). 114. Reynolds, P. W., and Chapman, D. I,., Proc. Roy. Soc. London A166,284 (1936). 115. Chapman, D. L., Ramsbottom, J. E., and Trotman, C. G., Proc. Roy. SOC.London A107, 92 (1925). 116. Benton, A. F., and Emmett, P. H., J . Am. Chem. Soc. 47,346 (1925). 117. Wieland, H., The Mechanism of Oxidation. Yale University Press, New Haven, 1932. 118. Warburg, O., Ergeb. Enzymforsch. 7 , 210 (1937). 119. Schlenk, F., A Symposium on Respiratory Enzymcs, University of Wisconsin Press, Madison, 1942. 120. Warburg, O., and Chrktiaii, W., Helv. Chim. A d a 19, E 79 (1936). 121. Hogness, T. R., A Symposium on Respiratory Enzymes. University of Wisconsin Prcss, hladison, 1942. 122. Kuhn, R.,and Strobele, R., Ber. 70B,753 (1937). 123. Michaelis, L., and Schubcrt, M. P., Chem. Revs. 22,437 (1938). 124. Stephenson, M., Bacterial Metabolism. Longmans, Green, New York, 1939. 125. Cavanagh, B., Horiuti, J., and Polanyi, M., Nature 133,797 (1934). 126. Farkas, A., Trans. Faraday SOC.82,929 (1936). a. Huberman, N. D., and Rittenberg, D., J. Biol. Chem. 147,211 (1943). 127. Taylor, H. S., Proc. Roy. Soe. London A108, 105 (1925). 128. Roginsky, S . Z . , J . Phys. Chcm. U.R.S.S. 16,1(1941). 129, Polanyi, M., J . SOC.Chem. Znd. 64,123T (1935).
CATALYTIC ACTIVATION OF HYDROGEN
130. Herington, E.F. G., and Rideal, E. K., Trans. Faraday SOC.40,505 (1944). 131. Stearn, A. S.,Ergeb. Enzynforsch. 7 , 1 (1938). 132. Quastel, J. H.,Trans. Faraday SOC.26,853 (1930). and Warhurst, E., Trans. Faraday SOC.34,614 (1938). 133. Evans, M.G., 134. Szent-Gyorgyi, A.,Nature 167, 875 (1946). 135. Lennard-Jones, J. E.,and Coulson, C. A., Trans. Faraday SOC.36,811 (1939).
199
This Page Intentionally Left Blank
Isomerization of Alkanes HERMAN PINES
Research and Development Laboralo+ies. Universal Oil Produck Company. Riverside. Illinois and Department of Chemisty.Northwestern University. Evanston. Illinois
CONTENTS I. Introduction . . . . . . . . . I1. Iaomerization of Butanes . . . . . . 1 Early Investigations . . . . . . . 2. High Vacuum Technique . . . . . . a . Effect of Promoters . . . . . . b . Effect of Olefins . . . . . . . c. Effect of Oxygen . . . . . . . d . Effect of Water . . . . . . . 3. Equilibrium Study of Butanes . . . . 4 . Mechanism of Isomerization . . . . . 5. Commercial Methods for Isomerization of n-Butane I11. Isomerization of Pentanes . . . . . . 1. Without Cracking Suppressors . . . . 2. Cracking Suppressors . . . . . . 3. Equilibrium Study of Pentanes . . . . 4 Mechanism of Isomerization . . . . . 5 . Commercial Methods for Isomerization of n-Pentsne IV. Isomerization of Hexanes . . . . . . 1. Without Cracking Suppressors . . . . 2. Cracking Suppreesom . . . . . . 3. Equilibrium Study of Hexanes . . . . 4 . Commercial Methods for Isomerizstion of Hexanes V. Isomerization of Heptanes and Higher Alkanes . 1. Without Cracking Suppressors . . . . 2. Cracking Suppressors . . . . . . References . . . . . . . . .
.
.
.
. . . . . . . . . . . . . . . . . . . . . . . . .
. .
. .
. .
.
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. .
. .
. .
. .
. . .
. . .
. . .
. . .
. . .
. .
. .
. .
. .
. .
.
.
.
.
.
Page 201 202 202 203 203 203 207 209 213 215 223 224 224 225 238 240 244
. .
. .
. .
. .
. .
244 244 247 247 252 262 252 254
.
.
.
.
.
255
. .
. .
. .
. .
. .
I INTRODUCTION The isomeriaation of n-bu tane and higher alkanes has assumed great importance in the last 10-15 years. The need for isobutane as a charging stock for alkylation processes and the necessity for converting low octane number straight chain hydrocarbons to their higher octanenumber branched chain isomers. gave an impetus to the study of isomeriaation of alkanes . 20 1
202
HERMAN PINES
The most important catalysts found for this reaction are aluminum chloride and aluminum bromide. The early investigators reported that these aluminum halides are catalysts for the isomerization of n-butane to isobutane; later investigators found, however, that the reaction proceeds only when traces of water or hydrogen halide are used in conjunction therewith. More recent work indicated that under certain controlled conditions, even aluminum halides-hydrogen halides do not catalyze the isomerization of butanes unless traces of o l e h s or their equivalent are present. The isomerization of liquid paraffins such as pentane and hexane proceeds also in the presence of aluminum halides-hydrogen halides. This isomerization, however, is always accompanied by decomposition. It was found that the decomposition can be inhibited by the presence of various substances such as hydrogen, cyclohexane, aromatic mono- and polynucleur hydrocarbons, isobutane, etc. In the case of heptanes and higher molecular weight alkanes the use of hydrogen was not successful in inhibiting the cracking reaction. Various other catalysts such as sulfuric acid, molybdenum oxide, etc., also have been investigated for the isomerization of alkanes. The isomerization in the presence of the oxide catalysts proceeds usually at high temperatures and in the presence of hydrogen. These catalysts have no significant importance. Sulfuric acid is not a general isomerization catalyst; it was successfully used in only a few isolated cases such as in the isomerization of Zmethylpentane to 3-methylpentane.
11. ISOMERIZATION OF BUTANES 1. Early Investigations The isomerization of n-butane to isobutane and the reverse isomerization proceed in the presence of catalysts containing either aluminum chloride or aluminum bromide; the latter, by virtue of its high solubility in hydrocarbons (Heldman and Thurmosd, 1; Boedeker and Oblad, 2) and its higher activity, causes the isomerization of butanes to proceed at lower temperatures. The early investigators of the isomerization of n-butane to isobutane did not take special precautions to purify the reagents used or to carry out the reaction under conditions which would exclude the contact of air, moisture, or other impurities. For this reason the results obtained are not always reproducible. a. Aluminum Bromide. Both n-butane and isobutane were isomerized in a sealed tube at about 27" in the presence of 18.9% by weight of aluminum bromide (Montgomery, McAteer, and Franke, 3). The isomerization proceeded rather slowly, and over two months were required to attain the
203
ISOMERIZATION O F ALKANES
equilibrium state (Section 11, 3) starting with either of the hydrocarbons. b. Aluminum Chloride. The early investigators of isomerization of hydrocarbons in the presence of aluminum chloride concluded that aluminum chloride is active as an isomerization catalyst only when used in the presence of hydrogen chloride. n-Butane a t a suitable temperature (Schuit, Hoog, and Verheus, 4) was converted in large yields to isobutane when contacted in an autoclave with aluminum chloride and hydrogen chloride. According to these authors the optimum temperature for isomerization was 100"; when the temperature was decreased, the rate of reaction dropped rapidly, whereas, when the temperature was raised, side reactions became pronounced (Table I). TABLE I
/
Egcct of Temperature an the Isomerisation of Buhnes i 7 ~the Presence of Aluminum Chloride-Hydrogen Chloride ~~~
Coinposition of Product Temp. "C.
Propane
Isobutane
%
%
20
...
40
...
60 80 100 120
...
15.0 17.6 22.6 35.1 62.9 16.7
... 3.8 70.5
n-Butane
Pentnnea and Higher
%
%
85.0 82.4 77.8 64.9 29.4 10.0
... ... ... ... 3.9 2.8 -.
The apparatus eonsirtcd of a 1-1. capacity stainlcss steel rotating autoclave. The reagents used in each experiment were composed of 80 g. butanes (11% isobutane and 89y0 n-butane), 40 g. aluminum chloride, and 8 g. hydrogen chloride. The duration of each experiment was 4 hours.
2. High Vacuu.m Technique
The isomerization of butanes using a high vacuum technique for the purification and introduction of the reactants mas investigated (Leighton and Heldman, 5 ) . It was found that completely anhydrous aluminum bromide did not act as a catalyst for the isomeriaation a t 85"; the addition, however, of a small amount of hydrogen bromide to the reactants caused n-butane to isomerize. a. Efect of P r O m O t e T S . The effect of various promoters upon the isomerization of butanes was also investigated (Heldman, 6). The experimental results summarized in Table I1 show that compounds such as sodium chloride, sodium bromide, boron fluoride, and the bromoparaffins are effective isomerization promoters for aluminum bromide. b. Efecl of OZefins. The isomerizstion of n-butane was studied with highly purified reagents using a high vacuum technique for charging and
204
HERMAN PINES
discharging the products (Pines and Wackher, 7). It was found that the isomerization of n-butane under relatively mild conditions did not proceed when catalyzed in the presence of either aluminum chloride-hydrogen chloride or aluminum bromide-hydrogen bromide. In order to cause such TABLE I1 The Effect o j Promotma upon the Isomeriaalion of n-Butane in the Presence of Aluminum Bromide _
_
~
Temp. 'C.
~
- ._
AIBrr, Molea..'100 Time. hm. moles n-butnne
Rind
Promoter Moles /lo0 niolen n-butane
-
Analysis % ' lsobutsne
25 25 25 84 *2 84 *2
96 72 149 12 12
10.4 29.5 3.7 2.9 2.4
HRr HBr HBr HBr HRr
13.5 7.6 4.0 0.7
1.7
14 17 38 8 15
84 *2 84 *2
58.7 44.4
29.G 6.5
HCI HC1
1.4 5.2
>0Q
25 25 *3 50*1 84+2 84 *2
96 286 48 59 44
17.1 6.1 4.0 1.6 3.0
NaBr NaBr NaRr NaBr NItBr
20.7 6.1 5.8 3.5 10.2
0.4 10 1.3 5 12
25 84 +2 M +2
96 44 35
34.7 3.3 2.8
NnCl NaCl NaCl
18.8 5.3 8.1
0.5 3840 31-36
25 84 *2 84+2
95 59 44
15.6 1.3 2.8
13Fn BFa BF,
1.8 5.0 10.6
>45
25 25 +3 84+2
96 405 12
10.6 7.6 17.5
CHIBr CHaBrl n-C.H,Br
16.0 60.0 6.1
11 75 38
59
20 42
isomerization it was necessary to add various compounds which are ordinarily present as impurities in the commercial butane isomerization. The addition of 0.01% butenes to n-butane was sufficient to cause isomerization in the presence of aluminum chloride-hydrogen chloride a t 100'. In the absence of o l e h s aluminum halide-hydrogen haIide did not cause the isomerization of n-butane, unless the experimental conditions were conducive to the formation of decomposition products. This was accomplished either by raising the temperature of reaction or by increasing the
ISOMERIZATION OF ALKANES
205
concentration of hydrogen halide. The data obtained by these authors were not in agreement with the experimental results described previously (Leighton and Heldman, 5 ) in which the butane probably was not purified sufficiently to remove components which might have acted as initiators for the isomerization. (1) Aluminum bromide-hydrogen bromide. The effect of olefins upon the isomerization of n-bu tane catalyzed by aliiminnm bromide-hydrogen bromide is summarized in Table 111. TABLE I11 Eflect of Olejins upon the Zsomerimlion of n-Butane Catalyzed by Aluminum Bromide-Hgdrogoi Bromide ~ _ - ~ _ _. _ _ . _ _ _ _ .... _ __ _ __ _ ~ ~ ~ . - _ _ _ _ _ _ ~ _ _ ~ ~
~
~
~
~
~
Experiment No. 1 2 3 4 Mole % butencs in n-biitrini? n 0.03 0.08 0.58 65.6 Isobutane produced, % 0.2 2.1 19.3 . .- ~.._____ ____ _ ~ _ _ _ Temperature of isonicrization: 25 =+=O.lo C. Duration of each experiment: 15 hours. Molal ratio of reagents used: ~ i - b i i t ~ n c~ ~ n l u m inum bromidefhydrogcn bromide= 100/9.3/2.3.
c -
~
The table s h o w that the addition of 3 parts butenes per 10,000 parts n-butane is sufficient to start the isomerization, while in the presence of 5.8 parts olefins per lo00 parts n-butane the yield of isobutane formed was very large. (2) Aluminum chloride-hydrogen chloride. Table IV represents the TABLE IV Efect of Olefina upon the Isomerizalion of n-Butane Catalyzed by Aluminum Chloride-Hydrogen Chloride ~.
Experiment No. Mole % butenes in n-butane Compdition of the hydrocarbon layer aftcr reaction, mole % n-Butane Isobutane Pentanes and higher Isobutane in the butane fraction, mole %
___ _.
1
2
0
0.013
99.9
87.6 11.8
CO.l
-0.1 0
3 0.05
0.6
87.0 12.6 0.4
11.9
12.7
_______
'
5
4 0.09
0.28
6 2.49
82.8 16.9 0.3
72.3 26.8 0.9
44.9 51.0 3.9
17.0
27.0
53.2
The molal equivalents of reagents used in each experiment: n-butane, 100 moles; aluminum chloride, 5.9 moles; hydrogen chloride, 3.1 molcs. The experiments were made by heating the reagents in sealed tubes for 12 hours fit 100".
206
HERMAN PINES
results obtained from the study of the effect of olefins upon the isomerization of n-butane catalyzed by aluminum chloride-hydrogen chloride. The reaction was carried out by heating n-butane, aluminum chloride, and hydrogen chloride in a sealed tube at 100" for 12 hours; the respective molal concentration of these reagents was 100/5.9/3.1. The concentration of butenes based on the charged butane varied from 0 to 2.6 mole %. It was found (Pines and Wackher, 7) that aluminum chloride-hydrogen chloride did not cause the isomerisation of n-butane to isobutane at 100". The presence however of 1 part butenes per 10,000 parts n-butane was sufficient to cause the isomerization. When the concentration of butenes was increased to 2.5% side reactions occurred, as evidenced by the formation of higher hydrocarbons and a small amount of a viscous dark layer; the latter was produced by a complex formation between the catalyst and unsaturated hydrocarbons. (a) Effect of temperature. The isomeriaation of n-butane proceeded also in the absence of added olefins when the temperature of the reaction washigh enough. This was probably due to the fact that at higher temperatures aluminum chloride-hydrogenchlorideacted also as a cracking catalyst, causing a partial decomposition of n-butane. The product of decomposition served as a promoter for the isomerization of n-butane to isobutane. The degree of isomerization, however, was greater when olefins were added. The experimental data are given in Table V. TABLE V The Effect of Temperature upon the Isonierization of n-Butane Catalyzed hg Aluminum Chloride-Hydrogen Chloride _-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __I_._____-~ ____-Experimmt No. 1 2 3 4 Temp., "C. 100 125 160 100 Reagents used: moles per 100 moles n-butane Aluminum chloride 8.91 9.26 9.50 9.21 Hydrogen chloride 6.87 3.24 3.19 6.86 Butencs ... . 0.09 A I ~ Y W Smolc , % Propane ... 0 3.3 0 0.1 16.3 62.4 16.9 Isobutrtne n-Butane 99.8 83.1 31.8 82.8 0.1 0.6 2.8 0.3 Pentancs
98
HC1 recovered, mole %
-~ -~-. ~ _ ~ ~ _ _ _ _ _ __.____I_.
-.
. ~
95
97 __
5 125 9.14 3.37 0.101
0 22.2 76.9 0.9
...
...
-__
.___ .___~
(b) Effect of hydrogen chloride. The isomerization of n-butane proceeded also in the absence of d d e d olefins when the hydrogen chloride
207
TSOMERIZATION OF ALKANES
concentration waO increased from about 6 t o 18 mole %, as shown in experiments 1 and 2 of Table VI. At more elevated temperatures, such as 150°, the concentration of hydrogen chloride had a pronounced effect upon the course of reaction of n-butane. In the presence of 3 mole % of hydrogen chloride, isomerization was the principal reaction, while in the presence of 7y0 hydrogen chloride, cracking occurred predominantly. TABLE VI Effect of Hydrogen Chloride Cmcenhatwn upon the Isonrerization of n-Butane _ _ ~ _ _ _ _ _ . _
Experiment No. 1 Temp., "C. 100 Reagents used: moles per 100 molcs of butme: 8.91 Aluminum chloride Hydrogen chloride 5.87 Analyses, mole % Propane Isobutarie 0.1 n-Butane 99.9 Pentanes arid higher 0.1 HC1 recovered, mole % 98 -
__
__
2 100
3 150
4
5
150
150
8.84 17.6
9.18 0
9.50 3.19
9.05 7.12
...
...
14.1 85.0 0.9
11.1 88.2
3.3 62.4 31.8 2.5 97
57.9 23.4 13.8 4.4 57 __ __
...
0.7
...
_____.
.. ~-
. .. -.
c. Efect of Oxygen. Since air is one of the constituents which might be present as an impurity during the isomeriaation of butanes, the effect of small amounts of oxygen upon the course of isomerisation was investigated (Pines and Wackher, 8). The catalyst used for this reaction wm either aluminum chloride or aluminum bromide. The former, only slightly soluble in the hydrocarbons, was deposited on charcoal in some experiments in order to increase the active surface of the catalyst. (1) Aluminum bromide. When aluminum bromide was used as a catalyst in the absence of air, n-butane did not isomeriae even when the solution was exposed to sunlight for 8 hours and then allowed to stand in the dark for an additional 500 hours. The isomerisation proceeded, however, when 0.06 mole air per 100 moles n-butane was added and the mixture was kept in the dark for 192 hours; the yield of isobutane produced was 7.5a/,. Nitrogen alone did not cause the isomerization under similar conditions. The degree of isomerisation is a function of the amount of oxygen added and of the length of time the material is exposed to sunlight. After 31 hours of exposure and in the presence of 0.055 mole oxygen, as air, per 100 moles butane, 63.6% isobutane was formed. Sunlight had a beneficial effect upon the rate of isomerisation. The experimental results are summarized in Table VII.
20s
HERMAN PINES
A quantitative study was made of the products resulting from the reaction of air with aluminum bromide dissolved in *butane in mold proportions: CIHlo/AIBrs/O1= 100/9.5/0.30. It was found that all of the oxygen entered the reaction; the gases which were pumped off, to mm. of mercury at liquid nitrogen temperature, consisted of 95% nitrogen and TABLE VII E#ed of Ozyyen uporb the Zeomerieation of n-Butane in the Presence of Aluminum Bromide at 86' - ~ .
__
Experiment No. 1 2 3 Duration of reaction, hrs. In dark 500 215 192 In ~iinliglit 8 ... ... Heagtmts used: molcs prr 100 molts dmtciiic Aluminum bromide 10.0 12.7 11.0 Oxygen 0 0.226a 0.06h Analysrs, mole % Isobutanc! 0 8.7 7.5 n-But aiir 99.6 90.7 91.8 ~-. ..
.~ ...
4
5
6
7
22 2
17 4
139 8
102
11.8 0.157
11.0 0.06b
11.1 0.044*
9.44 0.055b
15.7
0.5 90.1
29.7 59.0
63.6 35.0
84.3
31
Equal volunies of oxygen and air werc introdwed. Air wna used.
6% paraffinic hydrocarboIis. The condensable gases consisted of 63.6% isobiitane and 1.4% pentanes, the remainder being n-butane. Hydrogen bromide was recovered from the reaction, equivalent to 1.25 moles per mole of oxygen reacted. Part of the catalyst combined with hydrocarbons to form a complex which, after being freed of uncombined aluminum bromide, yielded unsaturated hydrocarbons on hydrolysis. The atomic rat,io of bromine to aluminum in the complex was equal to 2.5. The effect of oxygen upon the isomerixation of n-butane in the presence of aluminum bromide was also studied in a l-gallon glass lined steel bomb (Oblad and Gorin, 9). It was found that oxygen is an effective promoter in this type of apparatus also. It was noticed too that oxygen was used up during the reaction. This was in agreement with similar results recorded by others (Pines and Wackher, 8). The isomerixation followed a first order reaction with respect to n-butane concentration; the reaction rate at a given temperature was largely dependent on aluminum bromide concentration, oxygen concentration, and amount of surface. The rate of oxygen consumption was related to initial oxygen concentration, type of surface present, surface to volume ratio, and aluminum bromide concentration. The aluminum bromide consumption appeared to be related to oxygen consumption.
209
ISOMERIZATION OF A L K A N E S
It w a also ~ shown that bromine can be substituted for oxygen as a promoter (Oblad and Gorin, 9). (2) Aluminum chloride. Isomerization did not take place when n-butane was treated with aluminum chloride-charcoal catalyst alone. However, in the presence of a quantity of air such that the concentration of oxygen was approximately 0.01-0.02 moles oxygen per 100 moles n-butane introduced, isomerization t o isobutane t,ook place. The experimental data are given in Table VIII. TABLE VIII
‘
EJect of Oxygen upon the Zsomerizatiun of n-Butane Using Aluminum Chloride as a Catalyst d g6’ Experiment No. Duration of reaction, hrs. In dark In eunlight Agitation Reagents used: moles per 100 moles n-butane Aluminum chloride,” Oxygen (as air) Analyses, mole % Isobutane +Butane Pentanes and higher
1
2
3
121 0
141 0
0 16
Yes
YeS
No
16.07 0.0168
17.40 0.0196
17.05 0.0112
29.9 70.1 0
38.6 59.0 2.1
7.2 91.3 2.1
The reaction tubea contained 123 parta by weight of activated coconut charcoal, 10-12 meah, per 100
parta of aluminum chloride.
d. E$ect of Water. The fact that water promotes the isomerization of saturated hydrocarbons when either aluminum chloride or aluminum bromide is used as a catalyst, has been recognized (Nenitzescu and Cantuniari, 10). Avacuum technique was employed in the study of the effect of water upon the isomerization of n-butane (Wackher and Pines, 11). Two methods were used for this study: (a) Aluminum bromide or aluminum chloride was treated with various amounts of water, the mixture was heated, and any liberated hydrogen bromide or hydrogen chloride was pumped off. The resulting product, freed of noncombined hydrogen halide, WM then used aa a catalyst for isomerization of n-butane. (b) Aluminurn halide, n-butane, and water were mixed. (1) Aluminum bromide. It was found that the amount of hydrogen bromide liberated from the action of water upon aluminum bromide depended upon the molal ratio of water to aluminum bromide used. With the increase of the ratio of the water to aluminum bromide, the amount of hydrogen bromide liberated decreased. The maximum, 1.25 moles hydro-
TABLE IX Reaction of Aluminum Bromide u 4 h Water, and Zsomerization of n-Btctane in the Presence of the Resulting Reaclioli Produd
Experiment No.
Charge -4luminum bromide, moles Water, moles Moles H20/mole AIBrl RBr generated: moIes/moie AIBr3reacted Unreacted aluminum bromide, moles Unreacted water, moles
A 1 70 0.75
2 18 1.25
3 24 1.50
45 1.25
0.00849 0.01026 0.00856 0.02060 1.01 2.01 I .2.5 1.06 0.00103 0.00071 0 0
0.01027 0.03073 2.99 0.85 0 0
0.01 33 0.0532 4.01 0.69 0 0
4
5 20 1.50
6 112 0.50
0.0121 0.0727 6.00 0.01 7 0
0.01170 0.04672 3.99 0.67 0 0
0.00078
7 47 0.50 0.01177 0.07040 6.00 0.017 0
0.m1
2 5
z
B
ra
B Isomerization of n-butane in the presence of t h e catalyst described in A Reaction time, hrs. ' 20 20 Reaction temp., "C. 25.0 25.0 %Butane charged, moles 0.0935 0.1108 HBr generated, moles/mole AIBrl reacted in A 0.01 0.03 Analysis of hydrocarbon product, mole % Isobutane 18.7 30.9 n-Butane 81.3 69.0 Pentanes and higher 0 0.1
m
0.04
20 25.0 0.1428 0.01
20 25.0 0.1308 0
15 80 0.1256 0.23
12 80 0.1266 0
16.6 82.6 0.8
2.7 96.0 1.3
0.0 99.0 0.1
62.8 35.4 1.8
0 100 0
20 25.0 0.1105
21 1
ISOMERIZATION O F ALKANES
gen bromide produced per mole aluminum bromide reacted, was formed when the molal ratio of aluminum bromide to water was 1. When the water-aluminum bromide ratio was 2, 3, and 4, the respective amounts of hydrogen bromide produced were 1.1, 0.8,and 0.7 (Table IX, A). By using 6 moles water per mole aluminum bromide aluminum bromide hexahydrate was formed, and therefore no hydrogen bromide was evolved. The complex resulting from the action of water on aluminum bromide, freed from any noncombined hydrogen bromide, was used as a catalyst for the isomerization of n-butane. It was found that b y contacting nbutane at 25" for 20 hours with the catalysts, isomerization occurred when the molal ratio of water-aluminum bromide was 1, 2, or 3 (Table IX, B). When the ratio was 4, slight isomerization occurred a t 25") but appreciably more occurred at 80". With a ratio of 6, no isomerization occurred. No appreciable evolution of hydrogen bromide was noticed in these experiments. The results obtained demonstrate that the isomerizing catalyst formed by the action of water on aluminum bromide is not equivalent t o an aluminum bromide catalyst, inasmuch as the latter requires hydrogen bromide and traces of olefins to cause the isomerization of n-butane. The effect of water upon the isomerization of n-butane that contains aluminum bromide also has been studied. These experiments are not comparable with those described above, as the hydrogen bromide formed was not removed from the reaction zone. It waa found that aluminum bromide promoted by water, in spite of the free hydrogen bromide present, is a less active catalyst (Table X) than the corresponding catalyst described TABLE X Isomerizatim of n-Butane in the Presence of Aluminum Bromide and Water Experiment No. Duration of reaction, hrs. Temp., "C. Charge, moles per 100 molesn-butane: Aluminum bromide Water Moles HlO/mole AIBrl Hydrogen bromide generated per mole AIBrs Analysis of hydrocarbons, mole % Isobutane n-Butane Pentanes and higher
1
2 20 25
3 20 25 9.3 56.0
0.49
9.3 27.9 3 0.48
13.2 86.6 0.2
11.8 87.1 1.1
20 25 9.3 9.3
1
6 0.025
1.0 98.0 1.o
in Table IX. In this case, it was again shown that the addition of 6 moles water per mole aluminum bromide had no catalytic activity upon the isomerization of n-butane.
212
HERMAN PINES
(2) Aluminum chloride. The reaction of aluminum chloride with water is, to some extent, similar to the one of aluminum bromide; namely, the more water added, the less hydrogen chloride liberated. With a 1 molal ratio of water to aluminum chloride, 0.71 mole hydrogen chloride was formed; when the water-aluminum chloride ratio was increased to 2, only 0.29 mole hydrogen chloride was generated per mole aluminum chloride reacted (Table XI). The catalysts produced from the action of water on aluminum chloride were less active than those obtained from aluminum bromide and therefore the isomerization experiments were carried out at 80-100". Unlike aluminum chloride, catalysts produced by the action of water on aluminum chloride did not require the presence of olefins or hydrogen chloride to promote the isomerization of n-bu tane. TABLE XI Readwn o j Aluminum Chloride m'th Water and Zeomerizalwn of n-Butnne in fhe Presence of the ResuUing Reacfian Products
-_
-~
~
A Experiment No.
I
2 117 4.5
3 500 1
0.0116 0.0116 1.00
0.0113 0.0113 1.oo
0.0115 0.0230 2.00
0.0082 0.71
0.0033
0
0.0084 0.74 0
0
0
0
1
at 26" C. Duration of reaotion,hrs.: at 100" C. Charge Aluminum chloride, moles Water, moles Moles H,O/mole AlCI, Hydrogen chloride generated : Moles Moles/mole A1C18 Unreacted aluminum chloride, moles Unreacted water, moles
480
0.29
0
B IsomerBation of n-butane in the presence of the catalyats described in A 12 12 Duration of reaction, hrs. 80 100 Temp., "C. n-Butane charged, moles 0.125 0.124 H,C1generated,moles/mole AlCI, used in A 0.24 0.03 Analysh of hydrocarbon product, mole % 9.0 42.3 Isobutane 89.7 56.0 %Butane 1.3 1.7 Pentana and higher
12 80 0.124 0.46 14.7 84.6 0.8 ___ -
Isomerization of n-butane in the presence of aluminum chloride and. water waa also studied. The hydrocarbon and the water were added to the alumhum chloride and the reagents were heated for 12 hours at 100". The
ISOMERIZATION OF ALKANES
213
amount of hydrogen chloride produced in this reaction and the percentage of butane which underwent isomerization decreased with the increase of the molal ratio of water to aluminum chloride used. These experiments conFirmed also the observation made previously, that aluminum chloride to which 6 moIes of water was added showed no cataIytic activity toward isomerization (Table XII), TABLE XI1 lsonitrtzation of n-Butane in the Presence of Aluminum Chloride and Wader ___ Experiment No. 1 2 3 4 12 12 Duration of reaction, l m . 12 12 100 100 remp., "C. 100 100 Zharge! moles per 100 mole^ n-Butmw 9.0 9.0 Aluminum chlorido 9.0 9.0 Water 36.5 54.2 9.0 27.4 1 3 4 6 Moles HlO/mole AlC& 0.12 0.46 0.22 ICl generated, moles/mole AK 'I, 0.51 halysis of hydrocarbons, mole % 11.3 0 35.1 16.8 Isobutane 87.8 100 n-Butane 63.6 81.9 Pcntanes and higher 0.9 0 1.3 1.3 .~
~
~
....
.. __
...
_ _ _ I _
3. Equilibrium Study of Butanes a. Calculation of Equilibrium Concentrations. The isomerization of mtanes and higher paraffin hydrocarbons is a reversible reaction. The ?quilibrium of isomerisstion of the butanes, pentanes, hexanes, and hepLanes was calculated from the heat of isomerization at O'K., A H g O , for the iormal paraffin to the branched-chain isomer-and from the values of the 'ree energy function ( F o - H o o ) / T , for each isomer (Rossini, Prosen, and Pitzer, 12). For the isomerization of the normal paraffin to the isoparaffin n-CeHb(gW) =i-csHb(gtbs)
(1)
IFo is the standard free energy change at a given temperature, with each :omponent in its ides1 standard state of unit fugacity, and 3u t
md
AP" _ ?' --RLnK
rhere K is t,he equilibrium constant. Then
(4)
HERMAN PINES
214
For the isomerization reaction as given by eq. 1 K =fi,ffn
and, for low pressures K = pi/pn
where f represents the fugacity and p the partial pressure. The calculated equilibrium concentration of normal and isobutane is given in Fig. 1; the amount of the two isomers, in mole fractions, isplotted as a function of the temperature. b. Experimental Determination of Equilibrium Constant: The equilibrium constants for normal and isobutene were established by studying the isomerization of butanes from both the normal and isoparafh side (Pines, Kvetinskas, Kassel, and Ipatieff, 13). The isomerization was studied at 25,100, and 150" and under experimental conditions designed to reduce to a minimum the amount of secondary reactions. Depending upon the temperature level investigated, the experiments were carried out in sealed tubes, a stirring autoclave, or a continuous flow type apparatus. From the experimental results obtained, the concentration of isobutane in liquid and in vapor phase was calculated as a function of temperature. The results summarized in Table XI11 are in good agreement with equilibrium determination of other investigators (Moldavskii and Nizovkina, 14), (Montgomery, hlchteer, and Franke, 3), (Horne, 15). . TABLE XI11 Zsonterization of Butazre Mole % Ieobutnnc liuuid vnuor
Temp,
'C.
20 25 30 40 50 60 70 80 90 100 110 120 130 140 150
1.410 1.395 1.378 1.348 1.319 1.290 1.264 1.238 1.215 1.193 1.175 ...
1.439 1.426 1.412 1.389 1.368 1.350 1.334 1.319 1.306 1.204 1.282
...
...
... . . ~~~
.
81.7 80.9 80.0 78.4 76.8 75.2 73.6 72.1 70.7 60.2 67.8
86.3 85.5 84.7 83.0 81.3 79.6 77.9 76.2 74.5 72.9 71.2 69.6 68.0 86.5 65.0 .-
_ _ . -_. -
All tlhe values agree within a maximum deviation of 3% in composition
215
ISOMERIZATION OF ALKANES
with values based on the equation
RlnKa" = (2318/T)- 4.250,
where KO" is the vapor phase equilibrium constant a t zero pressure, and T is temperature expressed in OK. \
, < - o .
,
TEMPERATURE "C.
.9
;.e
5 .7
2 .6 .5 W
-I .4
0
I .3 2 .I .O
3 0 0 4 0 0 5 0 0 600700 TEMPERATURE OK.
800
9
0
0
~
FIQ.1. Equilibrium concentrations of the butanes.
4. Mechanism
of Isomerization
Various mechanisms have been suggested to explain the isomerization of butanes. a. Heldman Mechanism. This mechanism (Heldman, 16) is based primarily on the observation that (a) dry, pure aluminum bromide and aluminum chloride are not isomeristltion catalysts unless a hydrogen halide or substance capable of producing the same type of reaction with aluminum halide (e.g., water, alkyl halides, sodium halides, etc.) is present; (b) under mild conditions, isomerization is the only reaction observed with butanes. With the higher molecuIar weight paraffins, and with all paraffins a t higher temperatures, simultaneous disproportionation and "cracking" may also proceed. Heldman assumed that aluminum exhibits a coordination number greater than 4 during isomerization and considered AIX, a catalyst, which acts as both an acid and a base. He suggested that: (a) a t first there is an oriented collision of the catalyst with the paraffin. This produces a configuration in whicha hy drogen from the carbon atom [Cp,] in the butane chain is near a halogen of the catalyst and the methyl group [C,,,] is near the aluminum (Step 1, see below). (b) AIX4- acts as an acid toward the end methyl group and as a base toward the hydrogen. In this manner ti transient activated complex is formed with the methyl and hydrogen loosely attached to the catalyst. The hydrocarbon residue is restrained
216
HERMAN PINSS
from moving from the vicinity by the residual attraction of the :C& and H. (c) A shift then occurs of an electron pair from the middle to the new end carbon of the hydrocarbon fragment. (d) Concomitantly, the fragments rotate with respect to the catalyst, so that the latter is brought into a new position with the :CHaand H adjacent t o the middle and end carbon atoms respectively. (e) Reattachment yields isobutane as the product. According to Heldman’s mechanism, the isomerization of n-bntane proceeds as follows :
X
H
X
H C
H
H
X
4
H
CHa-CH-C:
:C
H
H
Step b
X XAlX
H
- C-CHI .
- C-CHI H
Step c
X X A1 X X CHI
CHI ..
H
- CH - C:H H
H
w ? Pd
Step e
b. Powell and Reid Mechanism. Tritium, the radioactive hydrogen isotope, Ha, waa utilized to determine the steps involved in the reversible isomerization of +butane to isobutane (Powell and Reid, 17). The catalyst used for this study consisted of aluminum chloride supported on alumina or charcoal. The isomerization experiments were carried out at atmospheric pressure with temperatures in the range of 107-128’. The experiments consisted in passing the following mixture of gases over the catalysts: (a) tritium hydride and butanes; (b) tritium hydride, hydrogen chloride, and butanes; (c) tritium butyl and hydrogen. It was found that exchange between hydrogen and tritium always took place, and that both 12- and isobutane contabed tritium. The degree of exchange was proportional to the degree of isomerization. The great extent of exchange with hydrogen chloride was attributed to its integral participation in the reaction where the exchange of hydrogen atoms hetsveen hntanes
ISOMERIZATION OF ALKANES
217
and molecular hydrogen is owing to the vulnerability to attack by hydrogen of the butane molecule in the process of isomerization. The following steps were proposed for the isomerization of n-butane to isobutane: "1. The butane contacts the catalyst mass in such a manner that the hydrogen of the hydrogen chloride in combination with aluminum chloride (designated as Ho in the diagram) enters the bonding sphere of the carbon atoms (designated as C,) of the butane. C,
Ci A1 Ci CI I10
/
}I
11
I
I
/
IIc'--C--C-
I
I
I1 H
I
/
CII
I 1
II
Ir
H II
(2)
(1)
(3) (4)
2. A weakening of the bond between C1 and Ca with a partial rupture occurs because-of the presence of three hydrogen atoms and two carbon atoms about the same bonding volume of C1. 3. At this point, the AlClr exists essentially minus its hydrogen in a position intermediate between C1 and Cs. In such a position, there is a tendency to acquire a hydrogen from Csthus leaving a situation of unfilled bond position distributed between Cl and Cs so that the methyl group Cz may either become bonded with Csto form isobutane or reattach itself to C, to return to the original normal butane form.l 1 According to the authors, the exchange of tritium from hydrogen to butane probably takes place at the point represented at the beginning of step 3. The reverse reaction, isobutane to n-butane is explained by the same steps in reverse order. The formation of hydrogen aluminum tetrachloride, as assumed in step 1, does not seem plausible in view of the reported recentwork (Brown and Pearsall, 18) which indicated that this compound is nonexistent. The interaction of aluminum chloride and hydrogen chloride does presumably occur in the presence of olefins. c. Chain Mechanism. (I) Effect of olefins. The mechanism described above (Section 11, 4a and 4b) did not take into consideration the fact that under controlled conditions n-butane does not undergo isomerization in the presence of aluminum hslide-hydrogen halide catalyst unless traces of olefins or their equivalent are present (Pines and Wackher, 7). In order to explain the role of olefins in isomerization, a chain mechanism was proposed
HERMAN PINES
218
(Bloch, Pines, and Schmerling, 19). This mechanism postulated that the isomeriaation of n-butane proceeds in the presence of aluminum halide and hydrogen halide by a sequence of reactions which may be expressed as follows in terms of carbonium ions.
+
It+ a. CH~CIT~CIT~CH~
4
+
CII&HCH&II,
+ RH
Reaction a serves only to initiate the chain and the reactant,.R+, need therefore be present in small amount only. The chain initiating ion, R+, may be produced in several ways. It may be formed in the presence of aluminum halide by the addition of hydrogen halide to an olefin which is present in the paraffin as an impurity, or which is added as such, or results from the cracking of the paraffin. I t may also be introduced in the form of an alkyl halide. The rearrangement of the s-butylcarbonium ion (b) finds analogy in the rearrangement of olefins in the presence of acidic substances (Whitmore, 20; Egloff, Hulla, and Komarewsky, 21). The reaction of a and c is similar to the hydrogen-halogen exchange reaction which occurs when isoparaffins are contacted with s- or t-alkyl halides in the presence of aluminum halide (Bartlett, Condon, and Schneider, 22). '
TABLE XIV Ismerimtion of n-Butane in the Presence of Aluminum Bromide-Deuten'um Bromide Catalyst -
.~-
Experiment No. Reaction time, liw. Temp., "C. Charge, mole
AIBr, DBr n-Butane Isobutanc Butenes Analyses of renctian product, molcs % DBr +HBr Iaobutane n-Butane Pentanes and highcr Per cent originally prcwnt as DBr found in butanes
__
2 20 25
1 20
25
3 20 25
0.01151 .01054 .1239 0 0
0.01021
7.9 0.0
6.3 93.3
9.4 36.0
0.4
63.0 1.o
9.6
92.0
92.1 0.0 6.0
)
4 7 25
0.01171 0.01031 0.1170 .01045 0 .I173 .lo80 0.1156 0 0 0 116x10-e 98x10-' .00942
8.1 19.1
72.4 0.4
84.8
219
ISOMEKIZATION O F ALKANES
In order to test the validity of the postulated mechanism of isomerization, deuterium bromide-aluminum bromide instead of hydrogen bromidcaluminum bromide waa used as an isomerization catalyst (Pines and Wackher, 23). The reactions were carried out under controlled conditions using a high-vacuum technique. The experimental conditions are summarized in Table XIV. It was found that by treating either n-butane or isobutane with 10 mole % deuterium bromide-aluminum bromide catalyst for 20 hours at 25", no isomerization of the butanes occurred and only 6 and 9.5% of the deuterium exchanged with n-butane and isobutane, respectively. When, however, 0.1 mole % butenes was added to n-butane and the isomerization reaction was carried out under the same experimental conditions, over 40% of the butane isomerized to isobutane and 92% of the deuterium underwent an exchange reaction. These results indicate clearly that olefins take an active part in isomerization. The results obtained are in agreement with the proposed mechanism of isomerization. The exchange reaction occurring in the presence of aluminum bromide between deuterium of the deuterium bromide and hydrogen of butanes, when small amounts of olefins are present, can be represented by the equations: CHS-CH=XXCHI
+ [CHSCHDCHCHJX+
+ DX + CHICHDCHXCHS+ +
CH&HDCHCHs F? CHSCHCHD
+
+ HX
(6)
CHsCCHiD
(7)
CHsCD =CHCHs
AH,
+ I
CHI
+
CHaCCH2D CHICH2CH2CH8 CHsCHCHID AH*
+
+ CH,CH2CHCHI
(8)
&HI
The deuterization of n-butane can also take place by applying the eame reasoning as given above. The equations also show the possibility of the existence of more than one atom of deuterium in a molecule of butane. (2) Effect of oxygen. Two hypotheses were proposed to explain the effect of oxygen upon the isomerization of n-butane in the presence of either aluminum chloride or aluminum bromide (Pines and Wackher, 8). Both hypotheses assume the formation of hydrogen halide and alkyl halide; the latter acting as a chain starter for isomerization. Hypothesis A. Aluminum halide reacts with oxygen to yield aluminum oxyhalide and halogen. The latter reacts with the alkanes to form alkyl halides and hydrogen halide. Alxs 1/2 On Alox + X, (9) C'Hio
+ * + Xs 3 CHs-CH-CH-CH,
I
+ HX
(10)
HERMAN PINES
220
The reaction represented in eq. (10) proceeds much faster in the presence of sunlight; the consumption of halogen causes a shift of'the equilibrium toward the right in eq. (9). The alkyl halide formed (eq. 10) acts, in the presence of catalysts, as a chain starter for the isomerization of alkanes.
Hypothesis B. Alkanes arc oxidized in the presence of aluminum halide to form AlXa
C,H,o
+ 0*-+2CdH,OH
(13)
The use of aluminum chloride as an oxidation catalyst was described in the literature (Friedel and Crafts, 24). The butanol formed (eq. 13) may react with aluminum halide to form butoxyaluminum dihalidc (Tsukervanik and Toltarcva, 25). C4HgOII
+ AlXs +CaHqOAIX, + €IX
(14)
Part of the butanol may also react with the hydrogen halide formed, according to the following equation: CIH90H
+ HX i=r C
(15)
The alkyl halide produced may act then aij a chain starter for the isomerization of butane as given in eq. (11) and (12). In order to throw some light upon the validity of hypothesis A, the effect of oxygen upon aluminum bromide was investigated. It was found that by treating aluminum bromide with pure oxygen a t room temperature, a slight but definite reaction occurred, as evidenced by coloration caused by liberated bromine. The product of such oxidation, after the removal of unreacted oxygen; showed slight catalytic properties toward the isomerization of n-butane, yielding 6.6-13.8% isobutane. The treatment of aluminum bromide at 100" with an approximately equal molal mixture of oxygen and nitrogen did not seem to increase the degree of oxidation. The experimental results are given in Table XV. The fact that the oxidation of aluminum bromide by pure oxygen or oxygen-nitrogen mixture is slow does
ISOMERIZATION OF ALKANES
221
not exclude the possibility that in hydrocarbon solution it may proceed much faster, especially since the liberated bromine reacts witahhydrocarbons and therefore is removed from the reaction zone. TABLE XV Effect of Prrtreatmenl of Aluminum Bromide with Oxygen upon the Isomerization of n-Butane
_.
.
Experiment No. Reaction time, hm. Temp., "C. Charged: Aluminum bromide, g. n-Butane, g. AlBr, moles/100 moles butam Hydrocarbons recovered. g. Analyses, mole % Isobutane n-Butane Pentanes and higher -
1 138 25
2 117 25
3 141 25
4 119 25
3.04a 7.07 9.32 6.90
3.1gb 7.45
2.24'
7.22
5.20 9.36 5.25
2.2& 6.34 7.38 6.23
9.2 89.5 1.3
6.2 93.5 0.3
6.6 92.3 1.1
13.8 86.0 0.2
...
a The aluminum bromide wae treated with 170 cc. (at s.t.p.) oxygen at 2&25" for 476 houm, aftar which the reaction v d was kept in liquid nitrogen and the oxygen was pumped off to 10 * mm. mercury preseure. b Treated with 170 cc. oxygen a t 20-26" for 683 hours: the reaction veasel wne kept at 0" while being pumped out. CThe aluminum bromidewas treated with 150 cc. oxygen-nitrogen mixture (46.3% 0s and 63.7%N~) at 100' for 1 hour. The vessel had a reddish tint because of the Liberated bromine. T h e gas88 were pumped out at room temperature; 140 cc. gas w w recovered. t i Treated with 135 cc. nitrogen-oxygen mixture (48.2% 01 .nd 61.8 N,) at 100" for 13 hours; 120 cc. gas consisting of 46.9% 01 and 53.1% Ns was pumped out at mom temperature and a t 10 8 mm. preeaure. From the analysb obtained it was calcubted that only 8 mole % aluminum bromide underwent oxidation to form aasumedly AlOBr.
(3) Effect of water. I t was found (Section 11, 2d) that by contacting 1 mole equivalent of water with 1 mole equivalent aluminum bromide (AlBrJ, 1 mole equivalent hydrogen bromide is liberated, while all of the aluminum bromide reacted (Wackher and Pines, 11). The reaction proceeded probably according to eq. (16). AIBrl
+ H 2 0+ HOAIBr2 + HRr
(16)
It was shown that hydroxyaluminum dibromide is an effective catalyst for the isomerization of n-butane even in the absence of hydrogen bromide and/or added olefins. In order to determine whether the halogen or the hydrogen of the hydroxyaluminum dibromide was the active part of the catalyst, deuteroxyaluminum dibromide was prepared by the action of deuterium oxide on aluminum bromide (Pines and Wackher, 23). D20+ AIBq +DOAlRr2 + DBr (17) It was found that by carrying out the isomerization in the presence of the deuteroxyaluminum dibromide and under essentially the same experimental
222
HERMAN PINES
conditions as for experiment 1, Table IX, 5.4% of deuterium underwent an exchange reaction although 21% isomerization of n-butane to isobutane occurred. The degree of hydrogen-deuterium exchange reaction is not proportional to isomerization; virtually the same percentage of exchange was noticed in a reaction in which the degree of isomerization of n-butane to isobutane was only 3.7%. TABLE XVI Reaction of Aluminum Bromide with Deuterium Oxide and Isomerizatia or ?&-Butanen the Presence of the Resulting Reaction Product
-
~ _ _ _ _ -
-
_____
--
-
_____.____
A
Experimimt No.
16 0.5
2 17 0.75
0.01175 0.01177 1.oo
0.01137 0.01142 1.00
1.oo
1.04
1
Reaction time, hrs... a t 25' at loo" Charge: Aluminum bromidc, moles Deuterium oxide, moles Moles D20per mole AIBr, Deuterium bromide generated Molm per mole AIBrS
B Isomerization of n-butane in the presence of the catalyst described in A. Isomerhation of n-butane in the prcscncc of DOAIBm. 43 20 Reaction t h e , hw. Reaction temp., "C. 25.0 25.0 0.1179 n-Butane charged, moles 0.1189 0.0 Analysis of product, mole % HBr and DBr 0.0 21.0 Isobutane 3.7 78.1 n-Butane 90.3 Pentanes and higher 0.0 0.3 -__
-
From the above it was concluded that the isomerization of n-butane by means of hydroxyaluminum dibromide involves the bromine atoms. The isomerization proceeds probably through a chain reaction. The chain is probably initiated by an exchange reaction as expressed in eq. (18). CHs-CH2-CH2-CIIs CI&--CH2-C~,II--CHs
+ DOAIBrl 7=t
+ [DOAIBrH]or (AlOnr + HD]
ISOMERIZATION OF ALKANES
CHs-CH-CHa
I
223
+ CH2-CHrCH-CHI
CHa
The steps expressed in eq. (19) and (20) are the same as described in section 11, 4c (1). Recent experiments (Beeck el al., 25a) on the isomerization of propanel-Cl* to propane-2-C1a further demonstrate that the isomerization proceeds through an intramolecular rearrangement. No propane containing more than one C1a carbon atom per molecule was found. These experiments were made by contacting vapors of propane a t approximately 25" and 450 mm. pressure with a catalyst prepared by the addition of 0.023 g. water to 0.40 g. anhydrous aluminum bromide for periods of time varying from 0 to 1074 hours. It was found that the rate of isomerization of pro~ane-1-C'~ to propane-2-U3 is comparable to the rate of isomerization of mbutane to isobutane under similar conditions. The equilibrium distribution of the propanes-l-C1a and -2-C1a was found to be statistical, that is, pr0pnne-l-Cl~/propane-2-C~~ = 2. The mechanism by which propane isomerizes can also be explained by means of a chain mechanism, in a similar way as was done for butanes. 6 . Commercial Methods for the Isomerization of n-Butane During World War 11, the great demand for aircraft fuel necessitated the production of large quantities of isobutane, a basic raw material in the production of high octane aviation gasoline. (See chapter on Alkylation of Alkanes.*) Various processes have been developed for the isomerization of n-butane to isobutane; all of them employed aluminum chloride-hydrogen chloride as catalyst. The difference between the various processes consisted either in the method of introduction of aluminum chloride to the reaction zone, the catalyst support, or the state of the catalyst. The following summary describes some of the main features of the various processes which were developed : a. The isomerization is carried out by mixing n-butane with a controlled quantity of hydrogen chloride and passing it at a temperature of 95-150" and under a pressure of 9-18 atmospheres over a catalyst composed of aluminum chloride supported on a granular alumina carrier (Cheney and Raymond, 26). The effluent from the reactor, after being cooled and condensed, enters an accumulator. Liquid from this accumulator is pumped to t h e hydrogen chloride stripping column where hydrogen chloride waa removed as the top product and is recycled t o thc feed strram. Isobutane is separated then from n-butane and the latter repassed through the catalyst. The average conversion of n-butane t o isobutane W D 45%. ~ b. This process consists in a continuous addition of aluminum chloride into the reaction zone, thereby maintaining thc activity of the catalyst constant (Chenicek el al.,
* Sttc Ipatieff and Schmerling, Advances i n Catalysis, vol. 11, in preparation.
224
HERMAN PINXS
27). This process has been dcvcloped by taking advantage of the fact that aluminuq chloride is slightly soluble in liquid n-butane. Thus a definite amount of aluminum chloride is continuously introduced into the reactor at constant rate by first pnssing all or a portion of the n-butane charge, in the absence of hydrogen chloride, through a heated zone called the saturator which contains granular aliiminum chloride. The quantity of aluminum chloride so introduced is accurately controlled by the temperature a t which the saturator is maintained and by the volume of butane passing through. The reactor in which the isomerizntion occurs is packed with quartz chips. The hydrocarbon stream from the saturator, containing aluminum chloride, enters the top section of the reactor and is mixed with another hydrocarbon stream containing the hydrogen chloride. The hydrocarbon effluent from the reaction zone which also contains hydrogen chloride and dissolved aluminum chloride, enters recovery sections. An aluminum chloride-rich stream is separated by distillation and recycled to the unit. Bubsequently hydrogen chloride is separated and recycled. Isobutane produced is separated from n-butane by fractional distillation. The isomerization is carried out at a temperature range of 50-120°, with the maximum conversion and greatest catalyst life at the lower end of the range. The isobutane content of the product k up to 50%. c. The catalyst used in thi3 process consists of anhydrous aluminum chloride dissolved in antimony trichloride together with hydrogen chloride (McAllister, Roes, Randlett, and Carlson, 28). The isomerization is carried out in a continuous manner employing continuous rejection of spent aluminum chloride and continuous catalyst make-up so that the catalyst is maintained at a.constant high level of activity. The conditions uRed for liquid phase butane isomerization arc: aluminum chloride content of the cetalyst 0-10%, hydrogen chloride content of feed 5% by weight, temperature of isomerization 80-90", and a prewurc of 20 atmospheres. Average contact time of the reaction is 12 minutes. The eonvcrsion per pass of normal to isobutane amounts t o 54%. d. The main feature of this process is to we a solid catalyst composed of aluminum chloride-bauxite and to maintain the activity of the catnlyst by the introduction of frmh aluminum chloride with the feed (Perry, 38). The dried butene is vaporized and superheated t o the desired temperature and passed on the rcactor which contains substantially dehydrated, low iron-content bauxite (Porocel). A controlled portion of the superheated vnpor is passed through a vessel charged with aluminum chloride prior t o entering the reactor. Aluminum chloride thus is vaporized and fed into the reactor aa required and is adsorbed on the bauxite. This technique enables the catalyst to be maintained in a relatively uniform and high state of activity. Typical reaction conditiom and results arc as follows: temperature, 120-150" C.; pressure, 14-17 atmospheres; feed rate, 0.5-1.0 volumes liquid butane per volume of bauxite per hour; conversion to isobutane, 40-50%; catalyst life, 180 gallons isobutnnc per pound of aluminum chloride, or 60 gallons isobutane per pound of bauxite.
111. ISOMERIZATION OF PENTANER 1 . Without cracking suppressors In the first systematic study of the isomerization of n-pentane it was found that freshly sublimed aluminum chloride does not catalyze the isomerization of n-pentane even on refluxing for several days (Glasebrook, Phillips, and Lovell, 29). When, however, a small amount of water, alkyl
225
ISOMERIZATION OF ALKANES
halide, or hydrated aluminum chloride was added to the reaction mixture, the n-pentane commenced t o undergo reaction almost immediately ; butanes and isopentane could be identified, the latter being the chief product. Substances boiling above n-pentane were also formed, but owing to the small amount present, could not be characterized. During the reaction the catalyst became coated with a layer of brown tar, which, after the hydrolysis, yielded a mixture of highly unsaturated, halogen-containing compounds. When gaseous n-pentane carried by a stream of nitrogen was passed over aluminum chloride at a temperature from 40 to 133", no reaction could be detected outside of a small discoloration of the aluminum chloride. The addition of hydrogen chloride caused the decomposition of n-pentane even at 40". Aluminum bromide proved a much more effective catalyst in causing the decomposition of n-pentane, and it was unnecessary knowingly to add other substances to the reaction mixture to induce reaction. It was found that n-pentane containing 1.5 mole % aluminum bromide underwent as much change as n-pentane containing 17.4 mole % ' aluminum chloride. The butanes formed increased practically linearly with the extent of the reaction, whereas the isopentane reached a maximum from which it slowly declined as the extent of the reaction was further increased. The effect of aluminum bromide concentration, reaction time, and promoters is given in Tables XVII, XVIII, and XIX. 2. Cracking suppressors
It has been shown that the isomerization of n-pentane in the presence of aluminum chloride or aluminum bromide containing catalysts was accompanied by side reactions, characterized by the formation of butanes and tarry material. These side reactions can be inhibited by the use of hydrogen or organic additives. TABLE XVII The Effect of Aluminum Bromide Concentrationon the Decomposition of n-Pentane Experiment No. n-Pentane used, g. Aluminum bromide, mole % Reaction loss, wt. % Products obtrtined: Butanes, volume % Isopentane, volume % Material boiling higher than n-Pentane, volume % The duration of the reaction waa 26.2 hours.
1 110 1.5 1.5
2 110 5.3 0.9
3 110 8.9 3.2
4 110 14.1 fi.8
1.5 27.2
6.1 53.5
10.2 46.8
14.1 40.4
2.8
3.1
5.4
5.9
226
HERMAN PINE6
TABLE XVIII The Effect of Reaction Time on the Decomposition of n-Pentane -___
-______.
Experiment No. 1 2 3 4 Reaction time, hrs. 22 51 95 169 Products ob hilied : Butancs, volumes yo 1.5 2.8 5.0 8.8 Isopentane, volume yo 40.8 51.0 55.9 44.2 Material boiling higher than n-pentane, volume % 2.5 3.4 3.4 5.9 One hundrcd and five g r a m of n-pcntanc was used in each experiment; the conccntration of nluxninum bromide waa 2.8 mole yo.
___
..
TABLE XIX The Effect o j Promoters in the Reaction of n-Pentane i n the Presence of Aluminum Bromide _
_
_
_
_
~
-
.
Experiment No.
.- -~ -~
-
1 With 2 ml. of water added
n-Pentane used, g. Aluminum bromide, mole % Reaction time Reaction loss and tar, wt. % Products obtained: Butanes, volume % Isopentane, volume % Material boiling higher than n-pentane, volume
200 3.5
3
2 Witlioot water
200
3.8 About 3 days 3.0 1.5
4
In the presence of hydrogen bromide
202 2.4
In the
absence of h dropn
crormde
202 2.4
About 38 hrs. negative
2.5
10.7 58.6
5.0 51.0
7.0 51.5
3.0 42.8
6.0
2.7
2.5
3.6
(a) Eflecl of Hydrogen. The use of hydrogen under pressure as a method for inhibiting side reaction during the isomerization of n-pentane has been reported by various investigators (Schuit, Hoog, and Verheus, 4), (Ipatieff and Schmerling, 30), (Evering et al., 31). A systematic study of the isomerization of n-pentane was carried out in a rotating autoclave of 850 ml. capacity (30). The catalyst and n-pentane were weighed into a glass liner equipped with a ground glasE capillary stopper. The open autoclave was flushed with nitrogen and the h e r was sealed into it. Most of the gas was then removed by applying suction for 2 minutes. Finally, a weighed amount of hydrogen chloride (if used) was pressed in and the pressure was raised to 100 atmospheret with hydrogen or nitrogen. The autoclave was then heated to the desirec temperature for 4 hours, after which it was permitted to set overnight. The experimental results obtained, summarized in Table XX, show tha-
TABLE XX
Experiment No. Aluminum chloride Gas
Temp., "C. Reactants, g. n-Pcntane Aluminum chloride Hydrogen chloride Recovered catalyst, g. Composition of product, mole yo Methane, ethane, propane Isobutane n-Butane Isopentane n-Pentane Hexanes and higher
Eflect of Hydrogen upon the Zeomerizalkm of n-Penlane 1
2
3 Resublimed
5
4
6
7 8 Technical HZ" Hz
9
Hz 125
H,
Hz
Ht
125
125
125
No" 125
125
100
Hz 125
100 10 0 16
100 10 0 10
100 10 2 10
100
100 12 1.6 12
100 10 0 15
100 10 0 13
100 10 0 10
12.9
0 0 0 2.8 96.6 0.6
0
0
4.0
0
0 22.4 77.3 0.3
4.6 47.4 2.8 30.6 3.3 11.3
4.2
0'5 0 61.0 38.0 0.5
27.6 18.4 5.5 25.4 10.2
10
10 10
" The initial pressure WRBB25 stmoepherea, in J1 the other experimemta the initial preaeure WIU
}
1.4 73.7 18.8 2.1
100 stmcepherea.
0 93.4 2.4 0
}
H2
150
g
363 35 0 37
100
2
5.8
20.8 1.4 2.5 65.2 10.1 0
0
3'2 14.2 82.6 0
U
r0 n
125
N?
10
0.1 84.9 9.2 0
10 0 10
P
g0
* !j q
1:
228
HERMAN PINES
little reaction occurred when n-pentane was heated with aluminum chloride and hydrogen at 125". When the reaction was carried out under nitrogen pressure, side reactions occurred; butanes, hexanes, and higher boiling alkanes were formed and the catalyst was converted to a viscous red liquid. Isomerization did take place under hydrogen pressure when hydrogen chloride was used as a promoter. The yield of isopentane increased with the amount of hydrogen chloride used. A yield of i'40/, isopentane was obtained when a small amount of water instead of hydrogen chloride was added to the reagents. The purity of aluminum chloride plays an important role in the activity of the catalyst. Technical aluminum chloride, containing 15 to 20% unsublimable product, was an excellent catalyst even in the absence of added hydrogen chloride; 85y0 of the n-pentane was converted to isopentane. On the other hand, isomerization of about one-third and autodestructive alkylation of almost all 01 the remaining two-thirds of the pentane charge occurred when nitrogen was substituted for hydrogen. The temperature of the reaction affects the degree of isomerization and decomposition; at loo", 14% and at 125O,85% of the n-pentane isomerized to isopentane; at 150", however, 2501, of the material recovered consisted of gaseous alkanes. The effect of various hydrogen pressures upon the inhibition of the disproportionation of pentanes was investigated (Evering et al., 31). The reaction was carried out in a 1490 ml. capacity reactor which was stirred mechanically. The reactor was jacketed so that it could be quickly heated by steam or cooled by water. Provisions were made for the withdrawal of the liquid product from the settled aluminum chloride catalyst. The experimental results obtained are summarized in Table XXI. TABLE XXI Egect of Hydrogen Pressure Upon Isme+ization of Pentanen
Experiment No. Hydrogen pressure, atm. Products formed: Weight % condensiblcs Per cent isopentane (basedon pentanes)
1 0.7
2 10.2
3 13.6
4 17.0
69.5 71
2.3 71
2.2
1.8 64
71
One liter of pentanes, composed of 12% isopentane and 88% n-pentane, 72 g. aluminum chloride, and 20.5 g. hydrogen chloride were used in each experiment. The temperature of the reaction was 100" and the duration 2 hours.
The results show that a minimum effective hydrogen pressure of 8.9 atmospheres exists at 100". This critical pressure, however, is dependent
ISOMERIZATION OF ALKANES
229
both on temperature and on catalyst activity. Increasing hydrogen presmre above this critical value retards isomerization, as well as effecting a slight additional decrease in the disproportionation reaction. b. E$ect of Organic Additives. The effect of organic additives in controlling side reactions in the isomerization of n-pentane t o isopentane in the presence of aluminum chloride-hydrogen chloride catalysts has been described. The organic additives best suited for decomposition inhibitors are mainly naphthenic and aromatic hydrocarbons and to some extent, isobutane. Batch and continuous types of experiments were carried out in order t o study the effect of additives upon the isomerisation of n-pentane (Maviky et al., 32). Batch type experiments were carried out in a turbomixer type autoclave. Relatively small amounts of material sufficed in this' type of test, which permitted a rather rapid survey of a wide variety of additives. The continuous tests were directed toward practical application of the method. A sketch of the equipment used for the continuous tests is shown in Fig. 2. The blend of pentane and additive was pumped a t a constant rate into the bottom of the saturator from a calibrated pressure charger by means of a high pressure differential plunger pump consisting of a steel tubing 91.5-cm. long and with a 2.5-cm. inside diameter. This was packed at the start of each run with granular, anhydrous aluminum chloride. The effluent from the saturator, consisting of a solution of aluminum chloride in the charging stock, was discharged to the top of the reactor. The reactor was packed in some instances with 0.6-cm. semiporcelain Berl saddles, in others with 3-8 mesh quartz rock chips. Effluent from the bottom of the reactor passed into a settling trap from which accumulated sludge, i.e., hydrocarbon-catalyst complex, was periodically discharged. The hydrocarbon-hydrogen chloride stream was released through an automatic pressure controI valve and was continuously washed in two glass towers containing water and caustic, respectively. The hydrocarbons were collected in receivers refrigerated by dry ice. Anhydrous hydrogen chloride was admitted at a definite rate at the top of the reactor. (1) Aromatic hydrocarbons. Benzene. An outstanding fact in connection with the use of aromatic hydrocarbons as cracking inhibitors is that the concentration required is very low. The criticality of aromatic concentration for the case of benzene is strikingly demonstrated in Fig. 3 agd Table XXII. The experiments on which the graph is based were carried out in the continuous type appahtus, at loo", a t a pressure of 7.2-30.4 atmospheres, and using 10 mole yo hydrogen chloride on hydro:arbons charged. In Fig. 3 both the degree of isomerization as indicated >y the isopentane concentration in the pentane fraction, and the extent
230
p3
WESfUR€
I N S U L A T E D TRANSFER TRANSFER LINE UNE INSULATED
lary
w
0
I
SOLENOID VALVES OIL BATH
7 THERMOCOUPLE
@--Y
NITROCCN
THERMOCOUPLE
HERMAN PINES
ULATION
I PRESSURE
GAGE
i
:HARGLA
TO
FIG.2.2. Continuous Continuousisomerizstion isomerizationunit. unit. FIG.
WASHING
TOWER
IBOMERIZATION OF ALKANES
231
of cracking as shown by the concentration of butanes in the product have been plotted against the percentage of benzene in the feed. The abscissa has been expanded t o show better the effect of variations in benzene concentration in the low range by using a scale in which the distance from the
60
R E A C T O R T E M P E R A T U R E 100. P R E S S U R E 7.2-30.4 A T M O S . L.H.S.U, 010
FIG.3. Benzene aa additive.
origin is proportional to log (ufl), where I ) is the volume yoof benzene in the feed. Optimum isomerization was obtained with about 0.25 to 0.5 volume % ' benzene. Thus, as increasing proportions of the additive were employed cracking was increasingly reduced. The isomerization reaction simultaneously became more,selective and reached a peak a t a point where cracking had become quite small. Further increase in additive concentration beyond this point gradually inhibited isomerization just as it inhibited cracking in the lower concentration range. I t was found that inhibition of isomerization by the higher benzene concentrations could be counteracted to some extent by a moderate increase in temperature but the optimum benzene concentration remained the same. The life of aluminum chloride catalyst with 0.5 volume yo benzene added to the pentane feed had been studied (Evering et al., 31). This study was made in an autoclave which was charged with aluminum chloride, pentanes, hydrogen chloride, and benzene. Agitation was started and then
h3 00
TABLE XXII
h3
BATCHTESTS
Eflect
01Organic A d d i t k s upon the Zsomerization of n-Peniane Mole % in product i-c&
n-CaHit
8.1
19.2
11.8
6.8
BENZEIW A N D n-ALKYLBENZENES 1.o . . . 0.9 ... 1.o ... 0.7 ... 1.7 .. . 0.0 . .. 1.7 ... 0.6 .. . 1.7 ... 0.0 ... 1.6 ... 0.3 .. .
48.7 50.7 7.2 2.0 7.0 0.8
49.7 48.0 91.9 96.9 84.3 93.9
0.7 0.6
Added compound Name
GlTkmS
None
...
Benzene Toluene Ethylbenzene n-Propylbenzene n-Butylbenzene n-Amylbenzene Isopropylbenzene see-Butylbenzene tert-Butylbenzene p-di-tertButy lbenzene
n-CBio
i-cbEtO
51.1
BRANCHED ALKYLBENZENES 1.7 . . . 2.3 .. . 1.5 2.0 1.6 1.7 . . . 0.0 . .. 1.7 3.0 0.2 POLYMETHYLBENZENES 1.7 . . . 1.3
...
CI
+
0.9
0.5 8.7 5.0
47.0 43.5 71.9 35.0
48.2 50.1 25.4 60.9
1.5 2.8 2.7 0.9
52.1
44.7
1.9
31.5 35.3 56.1 37.1
65.1 60.1 30.7 52.3
0.8 2.2 3.7 4.5
P O L Y C Y C ~ CA M ~ T I CHYDROCARBONS
Biphenyl Diphenylmethane Naphthalene %Methylnaphthalene
2.0 2.0 2.0 2.0
2.5 .
..
8.5 5.7
2.4
0.1 ._. 1.o 0.4
1
2
3m
Chlorobenaene flhlorotoluene o-Dichlorobenzene pDichlorobenzene Ijiphenyl ether Diphenyl ketone Aniline Diphenylamine
Gyclohexane Methylcyclohexane Ethylcyclohexane Dimethy lcyclohexanes 1sopropylcyc.lohexane 1,3- and 1,4-hfethylethylcyclohexmes n-Butylcyclohexane terl-Butylcy clohexane 1-Methyl-Pisopropylcyclohesanc Diethy Icy clohexanes 1-Ethyl-44sopropylcyclohexane
DERIVATIVES OF ABOMATIC HYDROCABBONS 1.o 3.3 0.6 2.1 25.3 0.9 1.7 50.7 7.7 2.0 49.1 5.4 . . . 1.4 ._. 2.1 2.0 43.2 2.2 2.4 43.7 2.3 2.0 49.4 4.7 NAPHTRENES 8.5 ... ... 8.5 ... 8.7 ... 8.6 8.6 8.7 8.8 8.1 8.1 8.8 8.8
.
... .. ,
. . . 5.0 . . . 2.8 . . . 12.0
.. . ... ...
35.0 14.4 22.3
1.6 6.8 10.2 9.4 0.7 7.4 13.5 2.3
19.6 26.3 53.6 23.2 26.7 25.3
32.1 9.0 11.8 9.8 44.3 24.0 13.8 18.3
45.5 55.4 57.3 57.2
46.5 35.6 40.2 39.7
6.4 6.5 2.5 2.7
51.2 .45.5 36.3 25.0 32.9 40.9 24.9
37.2 49.3 43.0 43.7 20.6
6.3 2.4 8.7 10.9
39.0
5.5 8.0
58.0
2
...
1.6 2.5 0.0 0.4
19.5
62.4
._.
0.8 2.6 0.2 2.8
0
42.0
8.9
5 E
B
8 b,
E 2
B
Materisls charged in each teat were 85 g. n-pentane, 15 g. anhydrous aluminum chloride, 2.7 +0.4 g. anhydrous hydrogen chloride and the indicated sdditive. Conditions were 6 hours at 75" C . Maximum preasure developed ranged from 3.3 to 8.2 atmospheres.
M w
TABLE XXIII CONTINUOUS TESTS E f m f oj Organic Addilbes upon the Zsomerization of Pentane Added Compound
Test
Vol. % in feed
Pentane feeda
hoursb
None
...
B
72-84
Benzene' Ben;enec Benzened Benzene Benzene'
o.io 0.25 0.50 2.0 10.0
Mole % in Product
penad
BENZENE D 79-1 87 D 72-204 D 24-68 D 244 D 48-84
+
~ C I H I O U-GHI,
iCsH11
. . .8.3.. .
li.3
67.9
6.3
. . .4.4. . . . . .2.3. . . . . .0.4. . . . . .0.3... . . .0.5. . .
33.6 47.3 49.0 25.8 15.9
57.2 47.2 48.1 69 5 70.1
4.8 3.2 2.6 4.5 13.5
. . .0.3. . . . . .3.9. . . .. .1.4.. .
54.7 56.8 54.3 53.3 60.6 49.4 44.2 32.6 19.2 4.5
31.4 30.1 30.1 32.0 21.5 45.5 51.1 60.0 67.1 80.9
13.6 9.2 14.2 12.6 13.2 4.9
nGHlt
C.
NAPHTHENES Methy lcyclopentane Cyclohexane Cyclohexane Cyclohexand Methy lcyclohexane Ethylcyclohexanec DimethgicyclohexanesC HydrindanC Decahydronapt halene Bicyclohexy 1
10.0 5.0 10.0 10.0 10.0 5.0 5.0
5.0 10.0 10.0
B B B A B C C C B B
48-60 6&72 48-60 72-84 2 4 4 120-168 36-132 36-120 91-107 48-60
1.3 0.8 . . -1.7.. . . . .0.3. . . . . .0.4. . .
. . .o.o.. . . . .o.o. . . . . .o.o. . .
4.3
7.5 13.7 14.6
4
1
2
cn
PARAFFINS Isobutanec IsobutaneC n-Hexane
10.0 20.0 20.0
C C B
3&168
48-108 24-36
14.7 1.2 26.2 1.2 . . .11.6. . .
30.3 31.2 17.3
47.0 35.3 50.4
6.8 6.1 20.7
Reactor conditions were unlesa otherwiae indicated in subaeqiient footnotea,34 atmcepherea, 100" C.. 0.10 LHSV (liquid hourly space velocity). H(:l ronrentration 10 mole 7~on the hydrocarbons charged. Saturator temperature was 77". a Commercial n-pentane feeds with the following compceitiona were used:
Feed A
B
C
D
* The Measured from start of indicated conditions. atmospheres. experiments were made at 17
The experiments were made at 10.2 atmospheres.
Composition, m d e % i-CjHir nCsFIii Cef 5.5 94.4 0.1 0.0 10.0 7.4
95.4 86.9 91.1
4.6 3.1 1.5
236
HERMAN PINES
thc reaction mixture was quickly brought up t o temperature. After a given time, agitation was stopped and the reaction mixture was quickly cooled. The aluminum chloride catalyst was allowed to settle and the bulk of the liquid product was withdrawn through a tube extending twothirds of the way to the bottom of the reactor. The reactor was then recharged and the procedure repeated a number of times.
Fro. 4. Effect of benscne on catalyst life in pentane isorneriztttion.
The results of these studies nre given in Figs. 4 and 5. The benzene inhibited the disproportionation reaction; only 4% butanes were produced whereas the isomerisation reaction proceeded readily. At a catalyst age of 212 moles pentane per mole aluminum chloride, the reaction product contained 66 volume yoisopentane. In the absence of benzene the isomerixation ceased after 55 moles pentane per mole aluminum chloride had been reacted. At a catalyst age of 185 moles pentancs per mole aluminum chloride the benzene was omitted. There was little disproportionation at this point because the benzene remaining in the reactor was dissolved in the catalyst complcx. However, in the subsequent run (also without benzene) a t 198 moles pentanr per mole aluminum chloride, a great deal of disproportionation was obtained, since a large proportion of benzene was extracted from the catalyst in a previous nm. The inclusion of 0.5 volume % benzene in the following experiment again inhibited the disproportionation. Substituted aromatics. A variety of different aromatic compounds
237
ISOMERIZATION OF ALKANES
wm investigated (Mavity et al., 32). The results obtained with each of these, classified according to type, are included in Table XXIII. The data are limited to a single concentration with each additive. Of the n-alkylbenzenes the effect of toluene equalled that of benzene. The higher alkylbenzenes inhibited both cracking and isomerization. The branched alkylbenzenes exhibited considerable variation in behavior. With tert-butylbenzene the effluent contained 71.9 mole yo isopentane, the highest concentration produced with any of the additives tested. Results with 1,4-dimethylbenzene equalled those with benzene or toluene . ~
11.5 WT:kON
C 1 2 WT.% O N C T E M PE R A T U R E
a W c 2
~
H E 4 2 V O L ‘ -1.90 P E N 1 . IG<-ALUU I I LOR I DE W E -HYDRO E N CHLORIO P
40-
-1
2 W
z
2
PO-
P
I
a5.k BENZENE
c:
I
0
1 -.
3
m
0
0 , ”
,,
0
0
I
30
I
MOLES PE N T A N E / W L E
D At C$
i
I
FIG.5 . Effect of benzene on disproportionation in pentane isomerization.
Several polycyclic aromatics exhibited positive inhibiting action. Aromatic hydrocarbon derivatives which were quite effective as inhibitors are chlorobenzene and diphenyl ether. (2) Naphthenes. Cyclohexane and the lower molecular weight monoand disubstituted cyclohexanes used in concentrations of about 5 to 10% in the feed blends were quite effective in repressing cracking in favor of isomerization both in batch tests (Table XXII) and continuous operation (Table XXIII). The higher molecular weight members of the series containing substituents with a total of four or more carbon atoms were less effective for inhibiting cracking. In the continuous tests certain bi- and dicyclics repressed both cracking and isomerization. The life of aluminum chloride catalyst in the presence of naphthenes
238
HERMAN PINES
was investigated under the same conditions used in life studies where aromatics were used 8s inhibitors (Evering et al., 31). The conversion to isopentane with 5.0yo cyclohexane was slightly lower than with 0.5y0 benzene, but the catalyst life was slightly longer. Decreasing the cyclohexane content from 5.0% t o 2.5'% at a catalyst age of 195 moles pentanes per mole aluminum chloride resulted in the doubling of the disproportionation side reaction. Careful fractionation carried out on samples (1 to 7 1.) of the effluent from the continuous runs with each naphthene and with benzene, showed that in all cases a major portion of the original additive was recovered as such or along with isomers. However, varying amounts of higher boiling products were usually recovered. With lower members of the nuphthene series and with benzene these products mere characterized as alkylnaphthenes and alkylbenzenes respectively. (3) Paraffins. In the presence of isobutane the extent of isomerization mas somewhat greater than when no additive was used but there was also a net production of disproportionation products (Mavity et al., 32). The effectiveness of isobutane in no way compares with that of cyclic hydrocarbons (Table XXII). The straight chain paraffin, n-hexane, had no apparent effect on the reaction. This might be anticipated from the fact that its homologue, n-pentane, does not suppress its own decomposition. n-Butane was likewise not effective as an inhibitor of disproportionstion. TEMPERATURE ' C , 27
127
227
327
427
527
627
727
TEMPERATURE IN OK.
FIG.6. Equilibrium concentrations of the pentanes.
3. Equilibrium Study of Pentanes a. Calculation vf Equilibrium Concentration. The cquilibrium constants of n-pentane, isopentane and 2,2dimethylpropane (neopentane), have
239
ISOMERIZATION OF ALKANZB
been calculated according to the method described in Section 11, 3a. (Rossini, Prosen, and Pitzer, 12). The concentration of the three pentanes as a function of temperature is given in Fig. 6. b. Experimental Determination of Equilibrium Concentration. The experimental determination of the equilibrium constants has been studied .by various investigators. The early study of the equilibrium constants of pentanes is open to criticism because the isomerization was accompanied by excessive decomposition (Moldavskii, Nizovkina, and Shterner, 33; Shuit, Hoog, and Verheus, 4). In order to establish the equilibrium constant, the isomerization from both the normal and the isopentane side was studied under experimental conditions designed to reduce to a minimum the amount of secondary reactions (Pines et al., 13). On the basis of experimental results obtained the concentration of isopentane and n-pentane in liquid and vapor phase was calculated as a function of temperature; the formation of neopentane TABLE XXIV Isomerization of Pentanea Temp.
"C.
60 70 80 90 100
110 120 130 140 1 50 160
Mole %, isopentane
Liquid
Vapor
87.G 86.8 86.1 55.3 84.6 83.9 83.2 82.5 81.9 81.3
89.6 88.6 88.1
...
170
...
1 80 190
...
200
. .
...
87.3
86.5 85.7 84.9 84.1 83.4 82.6 81.9 81.1 80.4 79.7 79.0
was not observed. The results are summarized in Table XXIV. calculated values were based on the equation
The
Rln K,.= 1861/~) -1.299,
where K p is the vapor phase equilibrium constant at zero pressure, T is the absolute temperature, and R is the gas constant. The value of 1861
240
HERMAN PINEB
cal./mole for the heat of isomerization of pentane isin good agreement with the value of 1930 cal./mole given by Rossini et al. (12). Evering and d'Ouville (34) have also studied the equilibrium of normal and isopentane in the presence of cracking suppressors. At 100" they found that the equilibrium mixture in the gas phase consists of 78.5% isopentane, while Pines et ul. (13) found the value of 81.9%. The reason for the difference between the two values is not known and further w-ork will be necessary to resolve this discrepancy.
4. Mechanism of Isomerization It has been shown that the isomerization of saturated hydrocarbons is affected by various factors inherent in the usual type of laboratory technique employed, and by impurities present in the commercially available reagents, (Pines and Wackher, 7, 8, 11, 23), (Pines, Abraham, and Ipatieff, 35). For this reason it is desirable for the study of the mechanism of the isomerization reaction, to deal with highly purified reagents; this is possible through the use of a high vacuum technique. The study of the isomerization of pentanes, although dealing chiefly with commercial grades of reagents or C.P. grades that were not rigorously purified brought out several factors which seem to throw some light on the mechanism of isomerization. The additives used for inhibiting the cleavage of pentanes during isomerization belong to either one or both of the three categories: (a)alkylatable inhibitors, such as aromatic hydrocarbons; (b) hydrogen donors such as decahydronaphthalene, or (c) molecular hydrogen itself. Compounds that can be alkylated and which are also known to act as hydrogen donors under certain conditions include isobutane, methylcyclohexane, and cyclohexane. In order to obtain some conclusive data as to the course of isomeriaation, several experiments were made using highly purified reagents and a high vacuum technique (Mavity et al., 32). These experimental data, summarized in Table XXV show that aluminum chloride-hydrogen chloride caused a great deal of decomposition of n-pentane. The addition of a little benzene inhibited both isomerization and the cracking of n-pentane. The effect of benzene was pronounced even when the concentration of hydrogen chloride was increased from 1 to 7%. These experiments indicate that benzene inhibits the cleavage of pentane; therefore carbonium ions, necessary for the isomerisation of n-pentane to isopentane are not produced in sufficient amounts for isomerization (Section 11, 4c(l)). The mechanism through which the inhibition a. Chain Mechanism.
ISOMERIZATIO~OF ALKANES
24 1
of the cleavage reaction proceeds is not entirely clear at present. Its complete clarification will require further experimental work carried out under strictly controlled conditions, including a study of the kinetics of the reaction. The addition of water to a mixture composed of pentane, aluminum chloride, hydrogen chloride, and benzene causes the isomerization of n-pentane. This is not surprising, inasmuch as it was already shown (Section II,2d) that hydroxyaluminum dichloride which is formed through the action of water on aluminum chloride is not equivalent to aluminum chloride-hydrogen chloride catalyst. The latter catalyst required an outside source of carbonium ions to catalyze the isomerization of n-butane to isobutane; hydroxyrtluminum chloride probably forms the carbonium ions necessary for this reaction through halogen-hydrogen interchange between the catalyst and the alkane (Section IIJ4c(3)). The commercial aluminum chloride may contain enough hydroxyaluminum chloride to cause such an isomerization without the introduction of carbonium ions from an outside source. This is brought out in Table XXV, experiment 7, where the commercial grade of aluminum chloride was used as a' catalyst. TABLE XXV Isomerization of n-Pentane Using High Vacuum Technique
Experiment No. Temp., "C. Time, hrs. Charge,"moles n-Pentane Aluminum chloride Hydrogen chloride
water Benzene Analysis, mole % Isobutane *Butane Isopenbe n-pentane Hexanw and higher
3 75 6
4 75 6
5 75 6
6 75
7 76
6
2 75 6
6
6
100
100
9.4 0
9.4 1.09
100 9.5 1.1
100 9.5 7.4
100 9.4 7.3
100 9.5 7.4
100 9.46 7.4
0 0
0 0
0 1.17
0 1.17
0.97 0
0.98 1.16
... ...
47.1 2.0 22.1 14.1 14.7
... .. . 0.3 , , .
... ... 0.1 ... ...
1
75
0 100
...
.. .
0 1.17
52.5 1 23.8 2.8 Trace 14.0 46.21 8.0 5.5 50.0 89.9 4.2 1.0 1.2
These materide were rieorously purified exoept aa indicated, in b. Commercial AlClr.
It was shown in Fig. 3 that with the increase in benzene concentration from 0.25 to 10% the degree of isomerization (including decomposition) decreases. In line with the experimental results obtained with the highly purified reagents, it is probable that in the presence of benzene the car-
242
HERMAN PINES
bonium ions necessary to cause the isomerization of n-pentane to isopentane were obtained through the presence of small amounts of extraneous material. By increasing the concentration of benzene the chance for the carbonium ions to react with it is greatly increased; for that reason the concentration of carbonium ions available for causing the isomerization is decreased. In order to obtain the same effect with naphthenes aa with benzene, the concentration of the former has to be much greater. The reason for this seems to be that the internction of a naphthene with a carbonium ion does not proceed as readily as with aromatic hydrocarbons. It is interesting to note that in the case of polyalkylated aromatics, or those which are not readily alkylated, the degree of decomposition is relatively high, which is in line with the data reported. 2,ZDimethylpropane (neopentane) which, according to the calculated equilibrium constants (Section 111, 3a) ought to be one of the main constituents of the isomerization of pentancs at room temperature, hae not been reported as present among the products of isomerization. This is not surprising in view of the proposed mechanism of isomerization (Section 11, 4c) which involves the formation of carbonium ions. In order for neopentane to be formed, it is necessary that a neopentyl carbonium ion be produced, which, through a hydrogen transfer reaction, would yield neopentane ;neopentyl carbonium ion, however, is very unstable and rearranges into a tertiary pentyl carbonium ion (Whitmore et al., 36, 37). The mechanism given above is mereIy a suggestion, based on presently available data, as to the function of the decomposition inhibitors in the isomerisation of n-pentane. It is hoped that a systematic study of the kinetics of isomerization of saturated liquid hydrocarbons in the presence and absence of cracking suppressors, may furnish additional information about the mechanism of isomerization. b. Mechanism of Evering, d’UuviEZe, Lien, and Waugh. This mechanism is similar to that postulated by Heldman (Section 11, 4a) and Powell and Reid (Section I114b)with the exception that the formation of free carbonium ion is postulated to lead to the rupture of the carbon chain as in cracking or disproportionation (Evering et al., 31). It is assumed that the actual catalyst in isomerisation with aluminum halide is a strong acid resulting from the association of hydrogen chloride and aluminum chloride [HCl. (AICIJZ]. It is postulated that the proton (H+) and an anion of thiE strong acid forms a loose association with a p a r f f i pictured below: If IT *. H:C:C:..
..
..
H H
(1) (2)
H +[CI(AlCl,)+]H.. H.. H .. C:C:C:II .. .. HHH
..
(3)
ISOMERIWTION OF ALKANES
243
The proton comes within collision distance of the C(a)atom of the pentane molecule. The net effect of the positive ion is to neutralize part of the electron negativity which leaves the Cp)atom with a somewhat more positive charge, which gives a driving force for shaking loose the C(1) atom minus its electrons in the form of a carbonium ion. The same action of the proton causes the hydrogen atom attached to the Cg)carbon atom to be loosened, minus its electron pair. This proton is attracted by the negative ion which is in a position close to the pentane molecule. With the pentane molecule activated in this manner, the driving force for rearrangement comes from the tendency to lose a proton from the Cg)carbon atom and a carbonium ion (CH3+) from the C(1, carbon atom. An intermolecular shift occurs whereby the carbonium ion migrates to the Cp) carbon atom and the proton from the acid becomes permanently bound to the Cp) carbon atom while the proton from the C(3) carbon atom goes to the anion, thus reforming from the strong acid catalyst H+ [Cl(AICls)z]-. As long as the catalyst and hydrocarbon molecule remain closely associated, no cracking occurs. However, when molecular hydrogen is split out and the carbonium ion moves farther than some critical distance from the catalyst environment, then cracking does occur. This carbonium ion can pick off a hydrogen atom from an adjacent hydrocarbon molecule, thus stabilizing itself and forming another carbonium ion t o propagate the chain; the carbonium ion can also lose a proton to form an olefin. Another carbonium ion can react with the olefin thus formed to give a larger carbonium ion, which should be very unstable and crack into an olefin and another carbonium ion to propagate the chain. Since the olefins formed by cracking are susceptible to carbonium ion formation owing to the acid catalyst medium, this leads to a rapidly increasing carbonium ion concentration. This results in an accelerating chain reaction which leads to the promotion of a large amount of cracked products, unless the chain reaction is broken in the early stages. According to this mechanism the cracking suppressors (Section 111, 2), such aa hydrogen or benzene react with carbonium ions before they set up a chain cracking reaction, thus stopping the chain. The mechanism described above does not explain the fact that under controlled conditions, benzene in the presence of aluminum chloridehydrogen chloride catalyst, inhibits not only the cracking but also the isomerization reaction (Table XXV, experiment 3) while in the absence of benzene cracking is the predominant reaction. The mechanism postulated above does not take into consideration the observations made that under controlled conditions saturated hydrocarbons such as methylcyclopentane, cyclohexane, or butanes (7, 23, 35) do not undergo isomerization, unless traces of olefins are present.
244
HERMAN PINEB
6. Commercial Methods for the Isomerization of n-Pentane
The following commercial methods were developed: a. This procws consists in paysing pentane containing 0.2 volume % benzene and hydrogen chloride over aluminum chloride deposited on Porocel (Perry, 38). The following are the typical operating conditions and reaults obtained. Reactor conditions : temperature, 83'; preesure, 17 atmospheres; feed rate, 1.0 volume liquid hydrooarbon feed/volume Porocel/hour; hydrogen chloride concentration, 1.0 weight % (2.0 mole %, based on feed. Reaulte: isopentane yield, 80%; selectivity, >99%; oatalyst life, 120 gallons of isopentane/pound aluminum chloride or >18 gallons isopentane/pound of porocel. b. This proccss consists in isomeriring pentmea and hexanes using a liquid aluminum chloride-hydrocarbon complex promoted with anhydrous hydrogen chloride. (Swearingen, Geakler, and Nysewander, 39). Hydrogen was added t o repress cracking and maintain high catalyst activity. The conversion waa accomplished by passing liquid hydrocarbon charge in a dispersed phase through the catalyst complex. The pressure, ; 48-55 atmospheres. Plants of preferred reaction temperature waa 115-1B0 5000 barrels per stream day capacity have been successfully operated, c. The catalyst employed is a molten salt mixture oonsisting of anhydrous aluminum chloride diesolved in antimony trichloride together with hydrogen chloride (McAllister et al., 28). The process is carried out in a continuous manner; the spent aluminum chloride is continuously rejected and continuous oatalyst make-up is employed in order to maintain the activity of the catalyst at a constant high level. The homeriation is carried out in the presence of hydrogen at a preeaure of.20 atmospheres and at a temperature of about 90-110'. The oonversion to isopptane per pass is of the order of So%, and thc consumption of catalyst is 1pound of aluminum chloride per 100 gallone.
IV. ISOMERIWTION OF HEXANES 1. Without Cracking Suppressors The action of aluminum halides on hexanes has been studied by many investigators (21, 40). It was reported that the reaction of aluminum chloride with n-hexane, promoted with water yielded a hydrocarbon layer boiling at 25-76', which appeared to contain some %methyl- and some 3-methylpentaneJ and a higher boiling fraction consisting of cyclohexane (Nenitzescu and Dragan, 41). T h e formation of the cyclohexane could not, however, be confirmed by subsequent work (Grummit et al., 4 2 ) ; it is likely that n-hexane contained some cyclohexane or methylcyclopentane. The product of the reaction of n-hexane with aluminum chloride and hydrogen chloride at reflux temperature consisted of isobutane and a mixture of paraffinic hydrocarbons boiling below and above n-hexane, of which one component appeared to be 2-methylpentane (Ipatieff and Grosse, 43).
The action of aluminum bromide on n-hexane using purified material and controlled conditions was studied (Gnunmit et al., 42). I n the absence of a promoter aluminum bromide does not homerim n-hexane at room
ISOMERIZATION OF ALKANES
245
temperature; this failure to react in the absence of a promoter is in agreement with previous work in which aluminum chloride was used as a catalyst (43). The effect of aluminum bromide-hydrogen bromide on n-hexane was studied by bubbling hydrogen bromide slowly through a solution of aluminum bromide in the n-hexane. The results obtained (Table XXVI) indicate that the isomerization of n-hexane is always accompanied by decomposition. The composition of the hydrocarbon layer was based on density, refractive index, and boiling point, and not on infraredanalysis; for that reason the composition of the hydrocarbon layer is open to criticism. TABLE XXVI
The Effect of Aluminum Bromide-Hydrogen Bromide upon n-Herane - _______. ~
~
Experiment No. Reagents used: n-Hexane, g. Aluminum bromide, g. Experimental conditions: Temp., "C. Time, hra.: Product obtained: In Dry Ice trap, g. Upper layer, g. Lower layer, g. Composition of upper layer, %h Isobutane Isopentane 2,2-Dimethylbutane 2,3-Dimethylbutane Zhlethylpentane 3-Methylpentane n-Hexane Heptanes Higher hydrocarbons I
1
2
310 51
272 74
Room 4
-20-0 143
25 263a 66 13.0 9.1 1.8 1.9 5.5 0.7 57.2 7.8 3.6
... 265" 62 4.3 5.2 9.6 3.8 10.4 0.6 64.5 1.1
After wsahin and drying. IncludeR the iquid in the Dry Ice trap.
The isomerization of branched chain hexanes in the presence of aluminum bromide-hydrogen bromide has been studied (Bishop, Burk, and Lankelma, 44). The summary of the results obtained, given in Table ,XXVII , shows that both 2-methylpentane and 3-methylpentane can iuomeriae with but a trace of cracking. The effect of reaction time on the conversion of 3-methylpentane shows two points of interest: the fivefold increase in cracking with the longer reaction time and the marked change in the relative concentration of the hexanes. It was found that 2, Mimethyl-
TABLE XXVII Efed of Aluminum Bromide-Hvdrogen Bromide upon Branched Chain H e z a n a 3xperiment No. &gents used: Hexanes, kind
Aluminum bromide, wt. hperimentdconditions:
1 %Methy lpent m e
%
Temp., "C. Time, hrs. knversion, % Zompositionof upper layer, % Is0butane Isopentane 2,2-Dimethylbutane 2,3-Dimethylbutane ZMethylpentane 3-Methylpentane Heptanes and higher a Unidentified hesanm 4.6%
6.3
2
3
2,2-Dimethylbutane 10.2 10.0
-11 217 39.2
0 137 82.8
1.1 1.4 1.4 10.0 60.8 17.6 7.7
6.8 23.3 11.5 7.2 17.2
1.8 8.0 79.5
8.4
...
25.6
6.1
0
149 20.5
a
...
4
5
3-hiethylpentane
6
7
2,3-Dimethylbutane
15.7
15.7
22.8
27.0
W
-11 337 83
-11 512
52.8
Z
90.2
-11 313 25
1.3 3.2 4.6 17.5 50.3 17.0 6.2
...
...
26.9 14.5 10.0 18.7 9.0 20.1
0.1 6.8 74.9 12.3
6.0
96 10.6 19.3 18.7 4.0 6.9 3.2 37.4
8
247
ISOMERIZATION OF ALKANES
butane is not appreciably isomerized. The composition of the liquid product was baaed on physical constants only, and not on infrared analysis. 2. Cracking Suppressors
It has been shown that the hexanes behave in many respects in a manner aimiIar to pentanes, namely the isomerisation is accompanied by decomposition, which can be inhibited by hydrogen and organic additives (Evering et al., 31). Thirty-four atmospheres hydrogen pressure was below the minimum effective pressure for inhibiting cracking in the isomerization of 2,3-dimethylbutane at 156". One per cent cyclopentane had no effect on the inhibition of side reactions in combination with hydrogen, whereas the addition of 10% cyclopentane to the 2,3dimethylbutane under the same conditions reduced the formation of butanes from 30 to 2.2%. The experimental results are given in Table XXVIII. TABLE XXVIII Effect of Hydrogen and Cyclopentane on the Conversion of 2,6Dimelhylbutane
Experiment No. Reagents used: 2,bDimethylbutme, ml. Cyclopentane, vol. yo Aluminum chloride, g. Hydrogen Chloride, g. Hydrogen, atm. Experimental conditions: Temp., "C. Contact time, min. Weight % of condensibla Analysie of liquid product, vol. 2.2-Dimethylbutme Pentanea 2,3-Dimethylbutane ZMethylpentane %Methylpentme n-Hexane and higher Cyclopentane
1
2
3
4
700
1000
990
50 15 68
73.6 20.5
34
1.0 69.0 20.5 34
890 10.0 69.0 20.5 34
100 20
156 10 32.6
157 10 30.0
160
8.9 41.5 4.5 14.4 7.3 22.5
... ... ...
21.6 5.1 6.3
...
37.8
... ... ...
11.8 11.3 6.0
*..
...
0.5 4.4 0.2
46.9 31.6 16.6
0.3
~
10 2.0
3. Equilibrium Study of Hexanes
a. Calculalion of Equilibrium Concentration. The equilibrium constants of the hexanes were calculated from the heat of combustion 8 8 described in section 11, 3a (Rossini, Prosen, and Pitzer, 12). The concentration of the five hexanes as a function of temperature is given in Fig. ?.
248
HERbUN PINES
b. Experinrsntal Determination of Equilibrium Cmtant. The experimental equilibrium constants for the isomeric hexanea were studied in an autoclave, which was mechanically stirred (Evering and dL’Ouville, 34). 27
127
TEMPERATURE IN ‘c. 227 327 427 527 627
727
z
0
l-
2
a LL
w
dI
,300 400 500 600
700 800 900 TEMPERATURE IN O K .
Id00
FIG.7. Equilibrium concentrations of the hexanca.
The hydrocarbon used in this investigation was a Mid-Continent naphtha having the composition given in Table XXIX. TABLE XXIX Composition uf Mid-Continent i,ight Kuphtha
_- _-
--
Isopentane n-Pentane Cyclopentane 2,2-Dimcthylbutane 2,3-Dimethylbutme 2-hlethylpcntane %Methylpentane n-Hexane C, naphthenes Benzene
Volume % 1.5 35.4 3.0 1 .o 1 .o 23.3 13.4 17.4 3.0 0.5
-
-
____
249
ISOMERIZATION OF A L a N E S
Although the hydrocarbon feed contained a considerable amount of material other than hexanes in the form of pentanes, naphthenes, and a small amount of aromatics, they were not believed to have any detrimental effect on the equilibrium of hexanes. I n fact, the naphthenes and benzene served the useful function of being decomposition inhibitors, thus reducing side reactions. The experiments were made as follows: 72 g. aluminum chloride or its equivalent aa aluminum chloride-hydrocarbon complex were placed in the reactor, usually with 1 1. naphtha, and sealed into the reactor. Hydrogen was added in the higher temperature runs until the total reactor pressure was 41 atmospheres at room temperature. In experiments at 21°, anhydrous aluminum bromide was used as the catalyst in an effort to establish the equilibrium more rapidly. The equilibrium constants for hexanes were determined at 21, 102, 161, and 204'. Since the contents of the reactor were cooled to room temperature before removal, the analyses reported in Table XXX are for the TABLE X X X Composition of Reactor Contenk at Equilibrium
Temp., "C. Moles Hydrogen Hydrogen chloride Propane 1sobuf;anr n-Butane Isopentime n-Pentane Cyclopentanc 2,2Dimethylbutane 2,3-Dimethylbutanc 2-Methylpcmtane 3-Methylpentanc n-Hexane hfethylcyelopent axle Cyclohexane Vapor: liquid ratio (at reactor tc~np.) Mole % decomposition
21
... .310 .010 .860
.044 2.600 .463 .209 2.295 .386 .879 .342 .167 .035 .573 liquid 5.0
\
102
101
...
.785 .493 .020 ,382 .232 2.515 .815 .271 .936 .346 1.341 .671 .419 .278 .272 .75 7.4
.350 .055 .920
,324 2.325 .414 .300 1.400 .367 1.005 .477 336 .173 .352 ,006 15.3
204
.900 .246 ,121 .714 .a3 1.841 .768 .131 .538
.no
.928 .440 .349 .200 .120 vapor 15.9 __
combination of the vapor and liquid phase which existed a t reaction conditions. The true equilibrium values for the hexanes in the vapor and liquid phase were calculated from these data by finding the vapor and liquid composition that existed in the reactor at the experimental temperature and pressure (Table XXXI).
% 0
TABLE XXXI Equilibrium Composition of the Isomeric H e z a i m RUll No.
Compdtion
1 2,%Dimethylbutane 2,tDimethylbutane 2-Methylpentane 3-Methylpentane n-Hexane
2 2,ZDimethylbutane 2.3-Dimethylbutane ZMethylpentane 3-Methylpentane n-Hexane
3 2,ZDimethylbutane 2,%DimethyIbutane %Methylpentane 3-Methylpentane n-Hexane 4 2,ZDimethylbutane 2,bDimethylbutane 2- hlethylpentane 3-Methylpentme n-Hexane
Temp. C.
Resetor pressure atmospheres
Vaporliquid =ti0
21
3.4
liquid
... ...
... ...
...
...
... ...
...
...
...
...
102
15.0 ...
0.006
...
... ... ...
... ...
...
161
59.5
0.75
...
...
...
... ...
...
...
. . .
...
...
...
...
204 ... ..
73.5
vapor
... ...
...
...
...
...
...
...
...
... ... ...
...
...
Observed
%%
Equilibrium Liqurd Vapor mole % mole 76
56.4 9.5 21.6 8.4 4.1
56.4 9.5 21.6 8.4 4.1
66.4 -82 16.9 6.0 2.5
39.1 10.2 28.0 13.3 9.4
39.1 10.2 28.0 13.3 9.4
45.0
25.2 9.3 36.1 18.1 11.3
24.9
25.9 9.7 35.5 17.8 11.1
21.7 8.9 37.5 17.8 14.1
9.1 36.4 18.2 11.4
... ...
. .
. . ...
11.7 7.2
21.7 8.9 37.5 17.8 14.1
1:
z
'd
251
ISOMERIZATION OF ALKANES
The agreement between the experimental vdues and those calcultbted from heats of combustion are good for n-hexane and 2,3-dimethylbutane but are in disagreement on 2,2dimethylbutane and the methylpentanes. The temperature coefficient for all the isomers is in good agreement but the TABLE XXXII Effect of Decomposition 022 Hezane Equilibrium Hydrocarbon Cyclopentane, vol. % Aluminum chloride, g. Hydrogen chloride, g. Contact time, min. Temp., "C. HZpressure, atm. Vol. %decomposition t o lighter than CI Analyses of Hexanea, vol. 7 0 2,ZDimethylbutane 2,3-Dimcthylbutanc 2-Methylpentane 3-Methylpcntanr n-Hvxane
9,,9-Dimethylbulane 8.3 69.0 20.5 10 160 37.5 7.4
2,3-Dimdhylbutans 0.0 73.6 20.5 10 156 37.5 62.9
22.6 8.8 37.3 18.3 13.0
21.8 11.0 35.2 17.8 14.2
TABLE XXXITA Scheme of Stepwise Isomerization of Hexanes n-Hexane F? 2-Methylpentane F? 3-Methylpentane
R 2,bDimethylbutane fsometization reaction o-Hexane -+ 2-methyl- and bmethylpentane ZMethylpentane + n-hexane SMethylpentane -+ bmethylpentane 3-Methylpentane -+ Zmethylpentane 2-Methylpentane 2,3-dimethylbutane 2,8Dimethylbutane -+ 2-methylpentane 2,3-Dimethylbutane +2,2dimethylbutane 2,2-Dimethylbutane -+ 2,3-dimethylbutane -+
2,2-Dimethylbutane Isomerization velocity IIKI7
1.8 0.7 150 65 6.0 3.9 0.8 0.04
heats of combustion predicted more 2,2dimethylbutane than has been Found experimentally. Aside from the absolute value there waa a good agreement in the ratio of 2-methylpentane to 3-methylpentane. The side reactions occurring during the isomerization of hexanes have no effect on the chemical equilibrium. This is shown in Table XXXII.
252
HERMAN PINES
A study of the behavior of the individual hexane isomem under controlled conditions was made wherein the per cent conversion waa varied over a wide range (Evering and Waugh, 44a). By plotting conversion against the per cent of the various hexane isomers in the product (based on 1 0 0 ~ converted) o and extrapolating back to zero per cent conversion an insight into the primary and secondary reaction products was obtained. From the experimental data it waa postulated that the hexanes isomerize in a stepwise manner according to the scheme in Table XXXIIa. The indicated isomerization steps do not take place with equal ease. Isomerization rates for the various hexanes to the specific isomers have been calculated in terms of the specific reaction velocity “K” based on the assumption that the isomerization follows a first-order reaction law.
4. Commercial Methods jor Isomwization o j Hexane Refer to section 111, 5b.
V, ISOMERIZATION OF HEPTANEB AND HIQHERALKANES 1. Without Cracking Suppressors The results of the many studies made on the action of aluminum halides on heptanes and their homologues indicate that the reaction requires a promoter such aa hydrogen halide and that it is accompanied by the formation of two layers; the upper layer consists of paraffinic hydrocarbons and the lower .layer of addition compounds between the aluminum halide and complex highly unsaturated hydrocarbons. The paraffinic hydrocarbons that result from this reaction contain a Iarge proportion of lower boiling hydrocarbons, the latter consists mainly of isobutane and isopentane. Extensive cracking occurred when aluminum chloride was added in small portions to a batch of heptane which was kept boiling, while the lower products of the reaction (which include any isomeric heptanes) were continuously removed through a fractionating ‘column (Calingaert and Beatty, 45). The identification of the hydrocarbons formed was based on the determination of the physical constants. An estimation of the products obtained is given in Table XXXIII, which shows that only 5% of the resulting hydrocarbons consisted of isomeric heptanes. The action of aluminum bromide on n-heptane, using highly purified materials and controlled conditions, has been studied (Grummitt et d.,42). Aluminum bromide, rather than aluminum chloride, was selected as a catalyst because its hydrocarbon solubility permits the initial reaction to be homogeneous. The reaction was carried out in a flask which was evacuated, flamed, and rinsed dry with oxygen-free nitrogen. The aluminum bromide
253
IBOMERIZATION OF ALKANES
was distilled into the reaction flask under diminished pressure. The hydrocarbon, after drying with phosphorous pentoxide, was distilled also into the reaction flask. The aluminum bromide dissolved in the hydrocarbon without sign of reaction. Hydrogen bromide was then bubbled TABLE XXXIII Products of Readion of Aluminum Chloride on n-Heptane
% PentaneE and below Hexanes 2,4-Dimethylpentane 2,2,3-Trimethylbutane 3,lDimethylpentane ZMethylhexane 3-Methylhexane Polymerization products
64.6 5.8
1.5 0.5 0.4
1.2 1.6
24.4
The yields were aaleiilatod on the basid of 100 parts n-heptane m e t e d .
slowly through the solution; as the reaction proceeded gas was liberated and a lower liquid layer formed. At the close of the reaction, the upper liquid layer was removed from the reaction flask by vacuum distillation. Table XXXLV summarizes the treatment of n-heptane with aluminum ,bromide. TABLE XXXIV Sumnuzry of Aluminum Bromide Treatment of n-Heptane -
Experimcnt No. Time, hrs. Temp., "C. Charge: mole % n-Heptane Aluminum bromide Hydrogen bromide Results : % Convewion to lower layer % Convcrsion,tot(a1
1 14.5 100
2 3.7 32
3 2.5 25
94.8 5.2
88.5
0
95.2 4.2 0.6
0.01 0.14
0.09 13.7
4
4 23
5 7.5
24
6 7.3 25
85.9 8.6
3.0
90.1 5.2 4.7
4.2 40.3
4.3 52.9
4.6
6.0
65
91.7
8.5
5.5
83.7 12.5 3.9
In the absence of a promoter, aluminum bromide had no appreciable effectupon n-heptane at temperatures up to the boiling point of n-heptane. Using hydrogen bromide as a promoter, reaction occurred readily at a temperature as low as -20". At room temperature and higher, cracking was the principal reaction. The conversion to lower boiling products was
254
HERMAN PINES
approximately from three to four times an great as the conversion to products boiling higher, which indicates that the principal reactions were cracking and isomerization. Of the two butanes, only isobutane was found in the products of reaction. Analysis of the liquid hydrocarbon fractions by distillation and physical constants showed that 2-methylhexane was present in the largest amount and 2,4dimethylpentane, 2-methylpentane, Zmethylbutane, 2,3dimethylbutane, 3-methylpentane1 2,2dimethylbutane, and 3-methylhexane were present in descending order of abundance. Although cycloparaffins have been reported (Nenitrescu and Dragan, 41) among the products of the action of aluminum chloride on n-heptane, subsequent work could not confirm this observation (42,45). 2. Cracking Suppressors The action of hydrogen upon n-heptane was studied in a rotating autoclave (Ipatieff and Schmerling, 30). It was found that hydrogen had a TABLE XXXV
.
Action of Aluminum Chloride on n-Hcptane .-
Experiment5 No. Atmosphere Reactants, g. n-Heptane Aluminum chlorideb Hydrogen chloride Pressure atm. at room tempcrature Initial Final Recovered catalyst, g. Composition of liquid mole % ' Propane Butanes Pentanes Hexanes Heptanee Octanes and higher Moles alkanes formed per mole heptane charged
1 N,
2 HI
100 10 0
103 10 0
102 10 0
101 10 6
102 10 5
100 100
100 100 15
100 100 10
100 75 15
100 100 17
...
20.6 47.5 21.3 8.3 2.3 0.0
7.2 12.0 9.0 6.2 62.1 3.6
1.48
0.94
18
3
4
5
H*
H 2
HI
3.2 18.0 20.8 0.5 50.6 6.4
61.4 6.5
21.2 15.2 11.1 45.0 7.5
1.10
1.05
1.01
2.5 16.6 11.3
1.7
5Experimenta carried out in glass liners in a rotating. 850-cc. autoclave. The autoclave waa heated at MO'for4 hours. bEeaublimedaluminum chloride wan used.
different effect on the action of aluminum chloride on heptanes and higher molecular weight hydrocarbons than on pentane and hexanes. With pure aluminum chloride, in the absence of added hydrogen, n-heptane underwent autodestructive aikylatiori at 150' regardless of whether the reaction was
ISOMERIZATION OF ALKANES
255
carried out under hydrogen or under nitrogen pressure. On the other hand in the presence of added hydrogen chloride and under hydrogen pressure, destructive hydrogenation occurred to give a high yield of lower boiling products, particularly isobutane ;under nitrogen pressure the heptane was autodestructively alkylated. When autodestructive alkylation occurred, the number of moles of alkane produced per mole of heptane was about 1.0-1.1; a substantial amount of octane and higher boiling material was produced. In the case of destructive alkylation very little product of higher molecular weight than heptane was produced, and the yield of alkanes per mole of heptane reacted was about 1.4-1.5 moles (Table XXXV). The use of hydrogen with heptane did not completely prevent the formation of a catalyst layer complex regardless of conditions, however, the complex which was obtained in its presence contained less hydrocarbon and was more active as indicated by the heat evolved by its reaction with water. The catalyst layer formed under nitrogen pressure was a brown tar. The action of aluminum chloride on higher molecular weight alkanes, e.g., those obtained from carbon monoxide hydrogen synthesis and boiling between 100-200", was analogous to its action on n-heptane. The foregoing study indicates that, unlike n-pentane and hexanes, hydrogen does not suppress the formation of side reactions during the action of aluminum chloride on n-heptane and higher alkanes.
REFERENCES 1. Heldman, J. D., and Thurrnond, C. D., J . Am. Chem. SOC.66,427 (1944). 2. Boedeker, E.R., and Oblad, A. G., J . A m . Chem. SOC.69,2036 (1947). 3. Montgomery, C. W., McAteer, J. H., and Frankc, N. W., J . A m . Chem. SOC.69,1768 (1937). 4. Sehuit, G. C. A., Hoog, H., and Verheus, J., Rec. truu. chin. 69,793 (1940). 5. Leighton, P. A.,and Heldman, J. D., J . A m . Chern. SOC.66,2276 (1943). J . A m . Chern. SOC.66, 1786 (1944). 6. Heldman, J. D., 7. Pines, H., and Wackher, R. C., J.A m . Chern. SOC.68,595 (1946). 8. Pines, H., and Wackher, R. C., J . A m . Chem. SOC.68,599 (1946). 9. Oblad, A. G.,and Gorin, hl. H., Znd. Eng. Chem. 38,822 (1946). 10. Nenitzescu, C. D., and Cantuniari, I. P., Ber. 66, 1097 (1933). 11. Wackher, R.C., andPines, H., J . A m . Chem. SOC.68,1642(1946). 12. Rossini, F.C., Prosen, E. J. R., and Pitzer, K. S., J. Research, Natl. Bur. Standards 27, 529 (1941). 13. Pines, H.,Kvetinskas, B., Kassel, L. S.,and Ipatieff, V. N., J . Am. Che& SOC.67, 631 (1945). 14. Moldavskii, B.,andNizovkina, T., J . Gen. Chem. U.S.S.R. 9,1652 (1940). 15. Horne, W. A., quoted in W. A. Cruse and D. R. Stevens, Chemical Technology of Petroleum. 2d Ed., McCraw-Hill, New York, 1942,p. 446. 16. Hcldman, J. D.,J. Am. Chem. SOC.66,1789 (1944). 17. Powell, T.hl., and Reid, E. B., J . A m . Chem. Soc. 67, 1020 (1945). 18. Brown, H. C., and Pearsall, H. Paper presented before the Division of Inorganic and Physical Chemistry, Am. Chem. SOC.(New York), Sept., 1947,
256
HERMAN PINES
19. Bloch, IF. S., Pines, H., and Schmerling, L., J . Am. Chem. Soc. 68, 153 (1946). 20. Whitmore, F. C., J. Am. Chem. Soc. 64,3274 (1932). 21. Egloff, G., Hulla, G. and Komarewsky, V. I., Isomerisation of Pure Hydrocarbons. Reinhold, New York, 1942. 22. Bartlett, P. D., Condon, F. E., and Schneider, A., J . Am. Chem. SOC.66,1531 (1944). 23. Pinea, H., and Wackher, R. C., J. Am. Chem. SOC.68,2518 (1946). 24. Friedel, C., and Crafta, J. hl., Com.pt.rend. 86,884 (1878). 25. Tzukervanik, I., andTokarcva, K., J. Gen.Chem. U.S.S.R.6,764 (1935). 25a. Beeck, O., Otvos, Y. W., Stevenson, D. P., and Wagner, C. D., J. C h .Phyu. 16, 255 (1948). 26. Cheney, €1. A., aud Raymond, C. L., Tram. Am. Imt. Chem. Engrs. 42,595 (1946).
27. Chenicek, J. A., Dryer, C. G., Sutherland, R. E., and Iveraon, J: O., Natl. Petroleam N ~ W36,40 S p. R-678 (Oct. 4,1944). 28. McAllister, 5.H., Ross, W. E., Randlett, H. E., and Carlson, G. J., Tram. Am. Inst. C h . Engru. 42, 33 (1946). 29. Glasebrook, A. L., Phillips, N. E., and Lovell, W. G., J. Am. Chem. SOC.68, 1944 (1936). 30. Ipatieff, V. N., and Schmerling, L. Papers presented before the Petroleum Division, Am. Chem. SOC., 111th meeting (Atlantic City), April, 1947, p. 265. 31. Evering, B. L., d’Ouville, E. L., Lien, A. P. and Waugh, R. C. Papers presented
before the Petroleum Division, Am. Chem. Soc., 111th meeting (Atlantic City), April, 1947, p. 285. 32. Msvity, J. M., Pines, H., Wackher, R. C., and Brooks, J. A. Papers presented before the Petroleum Division Am. Chem. SOC.,112th meeting (New York), Sept., 1947, p. 125. 33. Moldavskii, B., Nisovkina, I., and Shterner, R.A., J. Qen. Chem. U.S.S.R. 10, 1183 (1940).
34. Evering, B. L., and d’ouville, E. L. Papers presented before the Petroleum Division, Am. Chem. SOC.,110th meeting (Chicago), Sept., 1946, p. 259. 35. Pines, H., Abraham, B. M., and Ipatieff, V. N., J. Am.Chem. Soc. 70,1742 (1948). 36. Whitmorc, F. C., and Rothrock, H. S., J . Am. Chem. SOC.64,3431 (1932). 37. Whitmore, F. C., Whittle, E. L., and Popkin, A. H., J. Am. Chem. SOC.61, 1586(1939). 38. Perry, S. F., Tram. Am. Inat. Chem. Engrs. 42,639 (1946). 39. Swearingen, J. E., Geckler, R. D., and Nysewander, C. W., Tram. Am. Inst. C Engrs. 44,573 (1946).
h.
40. Thomaa, C. A., Anhydrous Aluminum Chloride in Organic Chemistry. Reinhold, New York, 1941. 41. Nenihescu, C. D. and Drggen, A., Ber. 66B,1892 (1933). 42. Gruinmitt, O., Seusel, E. E., Smith, W.R.,Burk, R. E., and Lankelma, H. P., J . Am. Chem. SOC.67,910 (1945). 43. Ipatieff, V. N., and G r o w , A. V., Znd, Eng.Chem. 28,461 (1936). 44. Bishop, J. W., Burk, R. E., and Lankelma, H. P., J. Am. Chem. SOC.67,914 (1945). 44s. Evering, B. L., and Waugh, R.C. Paper presented before the Petroleum Division of Am. Chem. h a . , 113th meeting (Chicago), April, 1948, p. 75. 45. Calingaert, G.,and Beitlty, H. A., J . Am. Chem. SOC.68,51 (1936).
The Application of X-Ray Diffraction to the Study of Solid Catalysts BY M. H. JELLINEK Linde A i r Products Company, Tonawanda, N . Y. AND
I. FANKUCHEN Polytechnic Institute of Brooklyn, Brooklyn, S. Y CONTENTS
PART I. DESCRIPTION OF METHODS I. Introduction . . . . . . . . . . 11. The Purpose of X-Ray Work 1. Identification . . . . . . 2. Orientation . . . . . . : 3. Angleof Scatter . . . . 111. Techniques . . . . . . . 1. Specimen Preparlrtioii . . . . 2. Cameras . . . 3. Monochromatom
.
.
.
.
P W .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
257 258 258 259 259
.
.
.
.
.
.
.
260
.
.
.
.
.
.
.
260 260 280 262 263 272
.
.
.
.
.
.
.
273 280 288
.
.
.
.
288
.
4. Average Crystallite Size . 6. Particle Size from Small Angle Sctltte&! .IV.Lattice Constants . . . . . .
?ART 11. EXAMPLES OF APPLICATIONS I. Identification . . . . . . . ‘11. Cry~talliteaand Particles . . . . 1x1. Conclusion . . . . ?ART 111. REFERENCES . . . .
.
PART I. DESCRIPTION OF METHODS I. INTRODUCTION In this discussion of x-rays and catalysis, no attempt will be made to eview critically the extensive literature dealing with the subject. Such xtensive reviews have been presented by others, notably by Prof. W. 0. dilligan at Gibson Island in 1944. Rather, a general discussion of the echniques of x-ray diffraction applicable to the study of catalysis will be #resented followed by a few specific appIications of these methods as arried out by the authors and their colleagues. Practically all x-ray scattering experiments are similar in principle. A 257
258
M. H. JELLINEK AND I. FANKUCHEN
welldefined beam of monochromatic or nearly monochromatic x-rays is allowed to fall on a specimen. A part of this radiation is scattered in all directions and is recorded by some device such as a photographic film, ionization chamber or Geiger counter. The pattern of scattered radiation is determined by the geometry of the experimental set up and by the scattering specimen. Because the structure of the specimen is one determining factor, one may be tempted to set up a one-to-one correspondence between the pattern recorded and the specimen structure. However, while the pattern of scattered x-rays can be predicted from the structure of the specimen, the revem proposition is not true. Only sometimes (rarely for complex structures) can the structure of the specimen be determined from the x-ray scattering. Nevertheless, much useful information can bc obtained from the x-ray diagrams.
11. THEPURPOSE OF X-RAYWORK Most of the diffraction work done in the field of catalysis has been dont using the powder method. This technique will be discussed from thret standpoints; namely, identification, orientation (principally electror dXraction) and angle of scatter. 1. Identification
(a) Identification by x-ray diffraction uses Bragg’s Law relating thc interplanar spacing in crystalline materials with the x-ray wavelengtl utilized and the sine of the angle of incidence of the x-ray beam and th+ crystal plane being considered. The significance of Bragg’s law is that b: suitably monochromatizing the x-rays the only two variables remainin) are the interplanar spacing and the angle of scatter. As the latter can b easily determined from the geometry of the camera or other recordin apparatus, the determination of the interplanar spacing is a simple a n t straightforward calculation. (b) The powder method has its greatest use because of the fact that theoretically, each crystalline material should give a unique diffractio pattern which would permit one to distinguish it from all other substancer Unfortunately, from a practical viewpoint, this uniqueness may be ver difficult to recognize, especially in the field of catalysts. Catalysts ar usually, but not always, finely divided substances, and accordingly yiel poor diffraction diagrams with fuzzy, smeared maxima. In fact, in many catalyst study, one can eliminate substances from further consideration L potential catalysts merely because they yield sharp diffraction diagrams c well-crystallized compounds. Because of this tendency toward poc diagrams, identifications may not always be clean-cut and unequivoca
APPLICATION O F X-RAY DIFFRACTION
259
Thus, many of the forms of gamma-alumina existing between the lowest temperature form and alpha-alumina are hard to characterize. Another effect that can be misleading in the study of catalysts consisting of various solid components is the masking of components even though physical mixtures in the same proportions clearly show all the components. Two cases that immediately come to mind are the complete absence of the lines of molybdenum trioxidc when it is present in quantities as high as 12% hy weight on gamma-alumina in a hydroforming catalyst, and the nondetection of gamma-alumina when adsorbed on the surface of silica gel until it comprises about one-quarter the weight of the catalyst. 2. Orientation When a specimen has no particular orientation (for example, a randomly oriented mass of small crystals, as in a powder), each reflection is spread out into a right, circular cone of radiation whose axis is the x-ray bcam. The intersections of these cones with the photographic films are the Debye-Scherrer powder lines. If, however, the crystals are not randomly distributed but lie in prcferred orientations, the powder rings become nonuniform in density, indicating orientation. One of the out.stmding studies of orientation in the field of catalysis was that of Beeck (1) who showed by electron diffraction that nickel films deposited under certain condit,ions showed unusual catalytic activity and that this activit,y was accompanied by a particular orientation of the nickel crystals.
3. Angle of Scatter I n the earlier applications of x-ray diffraction to catalyst study, the pattern of scattered radiation was determined for angles of 5" or more and the analyses consisted mainly of the usual Rragg's law application to determine t.he int.erplanar spacings of the samples under study. Scattering at small angles, i.e., st. angles less t,han about 2*, has been studied more recently following Guinier's (2) verification of Debye's (3) prediction that scattering of a cont.inuous nature should appear a t these angles for certain t.ypes of materials. This cont,inuous small angle scattering is different in character and in its theoret'ical interpretation from Hragg scattering. For the appearunc.e of scattering effects of this type it, is not necessary that any periodic st.ructure lie present in the specimen, nor is it w e n necessary that the mat'erial be cryst'alline. All that is required is that large scale discontinuit'ies be present in the materials such as discontinuities in the electron density (or refractive index for x-rays). These generally arise from discontinuities in densit'y. If regions of suit.uble size bounded by such dis: continuities are present, then a continuous small angle scattering will be ' observable. The intensity of the scattering, will among other things, de-
260
M. H . JELLTNEK .4ND I . FANKUCHEN
pend on the magnitude of the change in the refractive index across the discontinuities, while the rate of change with the angle of scattering will depend on the size and shape of the regions.
111. TECHNIQUES 1. S p c c i m n Preparation The preparation of specimens for Ikhye-Schemer powder diagrams has been adequately discussed in the literature and is fairly well standardized, coated fibers, extruded rods, or filled capillaries of glass, cellophane or of similar amorphous materials have been used (Bragg, 4; Jellinek and Fankuchen, 5 ) . 2. Cameras The cameras normally used for powder diffraction are also well standardized by the manufacturers of x-ray equipment and it is only an occasional special problem that will require an unusually designed camera such as an asymmetric focusing type or n special small angle scatter camera. Even these cameras are now commercialIy available. 8. Monochromators
To those familiar with powder work, monochromators usually represent a fine tool which is seldom called into use. The reason for the tendency to avoid crystal monochromatized radiation is that excessively lengthy exposure times have to be used. ‘l‘liis shortcoming however, is not always decisive and is often more than compensated for by the value of the results obtained. The very lengthy exposures encountered can frequently he greatly shortened by proper adjustment and camera design. (a) Plane crystal monochromators are usually made from cryst,als having a reflection with a large structure factor, i.e., high intensity, from one of its cleavage planes at a relatively smiill scattering angle to avoid polarization effects. The principle of operation is rather simple and may be described as follows: the radiation emitted by the x-ray tube target is usually a polychromatic band of racliation with a high relative intensity of the characteristic radiation of the target. A crystal, cleaved on a plane of known spacing, is then placed in this beam and is adjusted in such a position that the Rrngg angle for the characteristic radiation and the selected plane is established with the incidcnt beam. Under these conditions only the characteristic radiation, say the K alpha line, satisfies the requirements of Bragg’s law and is the only radiation reflected by the crystal. This reflected ray is thcn monochromatic except for the harmonics such as X/2 or X/3 which can usually be ignored or eIiminated by control of the tubepoten-
APPLICATION O F X-RAY DIFFRACTION
261
tial. If this beam is used to make diffraction diagrams, the resulting diagrams are clean with little of the confusing background usually found when white radiation is present. Simple monochromator designs have been published (Fanliuchen, 6 ; Lonsdnle, 7). Various crystals have been suggested as monochrornuting crystals and these have 1,ecn reviewed recently. (Iipson, Nelson, and Riley, 8). Fig. 1
FIG.I . Diffructioii tlirrgrams of the same iroit catalyst. In. Taken with filtered iron radiation. lb. Taken with iron radiation moiiocliromntizctl by a perita-erythritol crystal.
slioivs a comparison of two diagrams of the same iron catalyst taken with iron r d a t i o n . The upper pattern, la, was taken with filtered iron radiation and the l o w r , lb, was made using iron radiation monochromatized by a pentaeryt.hrito1 crystci.1. The improvement is obvious and monochromntizatiori is practically a necessity for careful study of these materials. (h) Bent crystal or focusing monochromators for x-rays have been known for some tirnc arid have the advantage of conserving the normally divergent radiation by focusing the beam to a sharp line, and thus of permitt,ing t,he use of shorter exposure times if cert'ain geometrical conditions are fulfilled. The coridit'ions are that the source of x-rays, the bent crystal, :ind thc point, of focus all lie on the circumference of :t circle (the focusing circle). The crystal is hent i1it.o a cylinder of radius twice the radius of the focusing circle anti then its inner surface is ground to t'he radius of the focusing circle to satisfy the focusing conditions and Uragg's law. Monochroniat.ors of this t . y p h \ ~ ebeen nrlequntely described. (Cuinier, 2; Johannson, 9; Hozorth and Hn\vurt,h, 10; DriRIond and IGrkpatrick, 11; Smith, 12.) Fig. 2 illuat~~ites the method of achieving t,he focusing effect by use of a bent c:iystJ ant1 Fig. 3 slio\~sone of thr \ ~ a y sin which i t was used by Gninier. Fig. 4 stio\w n st1rit.s o f diugixns obtninrtl Ily Guinicr wit.11 this apparatus using normal csposrtre times. This method not only requires
exposure times which arc! not longer than thosc normally required with filtered rudintion hut also yields diagrams which, due to t’hcir vcry low backgrounds, arc! far superior t.o any obt,:iinctl \vit,h filtered radiation.
4. Average Crystall.ite S.izu
FIG.2. Bciit crystal focusing.
used is tlint, givcn by Urtigg
wit,li
(a) Ordinary po\vder (DebyeS(*lierrcr)dia.gr:tms can he mc:tsurccl si.) :is t’o yield iiiform:ition cwiicvwiiiig t.he avcr:i:c crystnl1it.e size of’ tlic wnplr. ‘I’tiis iiiforni:it,ion is ol)t.nincd lyv nikasriring t,lw widths of t.hc cliffniciion lines tliemselves xi; their point of h d f m a s i m u m int’cnsity. The “half Imwltli” c;in tlicn he used to cal(w1ut’c the average crystallite siac. Formulas for muking - this cslculntion have been published by several workers but the one most gcncrally u nonwticin owing t o Warren.
L = 0.SSX B C C W8
where I, is tlic crystiillitc ili:riiic!tcr i i i A. X is t.ho wavi!lviigtli of x-ray mdintion i i i A. e is tlia Bmgg nnglc, i.c., anc-liiilf t.lic! angle of scsttcr. BCis thc corrcctctl width of line nt half mrisimuin iritciisit’p.
where
Strictly, this formuln applies only when several stringciit; condit.ioris itre fulfilled. The radiation used should be monochromatic a.nd the lines well resolved. I t also assumes strain free crystnls in t)he form of perfect cubelets all of about t,he same size. It shoriltl h poiii t.cd o i i i t.hnt, t,wo Inct.linrls : ~ r ccurrrntly available coiriinercially for rcccirding :\nd Iiic:isiit*ingtho irit’crisitirs :LOI’OSS the widths of x-ray ditfructiori Iincs. ’I’he first consists of registration on a photo-
APPLICATION OF X-R.4Y DIFFRACTION
263
graphic film followed by microphotometering. The second uses the x-ray spectrometer. The former inetliod requires n relatively expensive micro-
photometer :tiid tlic latter usiinlly gives ahorm:il line shitpcs clue to certain features of the apparatus. In both instnnws, it is necessary to run as a standard a wcll-crystallizer1 specinicn to obtnin the correction fartor of the particular apparntus used. Furthermore, \\hen :I \vide range of crystnllite
sizes is present in the solid under investig:itiori the theory is riot directly applicable. 5 . Particle Size jrorn Small Angle Scattering l‘hc ‘paiMe’ is one continuous piece of the material regardless of thc orientation of tlie crystallites of which it may be composed. Thus, a particle can never be smitller than one crystitllite hiit may be composed of one or more. As this field of x-ray scattering measurements is not LLSgenerally lie11 known and used as some of the others, it I\ ill be more completely discussed.
2G4
M . H. JELLlNICK A N D I. FANKll(!HUN
a. Gelaerul. Titeorg. The phenomenon of small angle scat.tering, after its prediction by Ilebye (3), for siihstances having tliscontinuit~ies,\\'as observed by Chinicr who made phot.ographs of this "central blackening" using a bent crystal monochromat.or uf t.he t,ypealready dcsrribctl. (iuinier (2, 13) also publishcd miwh of the theory. I t h:ts I>ten shown (Guinier, 2, 13; IIoscm:i.rin, 14; Jellinek :tnd E'anlaichen, 5; Iir:rt.liy, Sekora, and Tree,r, 15) t>hat for u homogenous assemblage, of spherical particles o f radius R , t.hc s m J l angle scattering of w a y s of wavelength X can be fairly accurittcly given by p?!??
I(Sl =C'H'/t%
3-
(2)
where 1 ( 0 ) is t,lw intmsity
:M
OF sc.st.tt.ring2.sR func.t.ioii OF e
is tlic mdss of thc scnttcviiig sptviiiii(*ii
c is a constnnt for a givcm iiltdcrinl For this case, a plot of log f (0) against f P should give a straight line whose slope will be :I funct.ion of 12. Thus for a preparation consisting of particles of :~pprosim.zt.elythe s:tm~'size (in the langc c : ~50-1OOO iI.diameter) a study of the small angle sc!att.cring of x-rays shoiild permit tho determination of the particle size. The discussion to this point' lias assumed that particles of only one size are prcscnt., which will seldom, if ever, be the caw in a c n td practice. When a weight dist,ribution of sizes esists, \shich is given by W ( R ) ,where W ( R ) is thc weight fr:tction of r:idius K,t'hen t'he curve of scattered radiation ns a function of thet.a is givcii hy
Now, t.he plot of log I against k2 or 8' is no longer a straight. line. In practice, the scattering curve is determined experiment.ally and U'(R) is the desired result. T'arious workers have stio\vn both analytically (Baiter, 16; Roess, 17) and graphically (Jellinek, Solomoil, and Fanliuchen, 18; Roerss and Shull, 19) that a particle size dist,rihiit,ion can be obtained from the experimental data. While thc analytic solutions \vould be more exact, the t,heory at, present is st.ill in nott t,oo gout1 II state and has bccn used in cases for which t Irr nssnmptions uridcrlying t'he theory do olwiously not apply. For t.hese rea.sons it. is felt, t.tiat the gmphical analysis is preferable and adeqr1atc at. presclnt. b. l'echniqiies of Measuring S.mall -4 r q l r Stutter. The osual small angle :tp,puntt.tis consist.s of some type of collimating system of considerable lengt,h t,o keep down to :L minimum the divergence in t.he beam, sometimes
APPLICATION OF X-RAY DIFFRACTION
265
combined with crystal monochromatization. (Alternatively a bent crystal focusing monochromator can b r used.) The beam is then passed through the specimen and the scattering is recorded, usually a t a fairly large distance from the specimen, especially when bent crystal monochromatization is not used.
FIG.5. Photographic record of sins11 angle scatter showing the changes produced by heating gamma-alumina.
(1) Photographic recording requires that a cassette be provided t o hold the film perpendicular to the x-ray beam. A stop, usually of lead or silver, is located a short distance in front of this cassette to catch and absorb the direct beam from the collimator to prevent it from halrtting the
111. H. JELLINEK AND I. FANKUCHEN
266
film and thus destroying the small angle region. Fig. 5 shows a series of small angle diagrams taken a t 600-mm. distance between specimen and plate, of a specimen of heated active alumina. (The larger the number in the diagram, the more severc the heating.) Filtered copper ridintion was used. The charigc in the size of the small angle scatter can be noted, especially in t'he M t e r port.ion of thc second column. 1nt.ensity measurements must be made over an intensit.y range of several thousand to one. This is a difficult task if phot,ogrsphic recording is used. For this reason, resort, has been made t,o Ckiger counters for recording t,he x-ray intensit,ies. Fig. 6
Fie;. 6.
Small l~iiglescatter ctpprtmtus w i t h Ckigrr couiiter.
shows such an apparatus (Jellinek, Solomon, and Fankiichen, 18). The scanning unit was mounted on pivots that allowed it to be rotated about the point where the x-ray beam and the specimen intersected. hleasurements made with this apparatus and with the photographic apparatus are compared in Fig. 7. In this figure, the intensity scale is relative and it is only the shapes of the curves that should be compared. It can be clearly seen that the apeement is good except in the high intensity region where a deviation is expected and encountered due to the poor photographic response of the photographic film in this region. Another important advan-
A1’l’LICATION OF X-RAY DIFFRACTION
267
tage of the Geiger counter apparatus is its higher sensitivity as compared with film. Figure 7 shows the Gcigcr ~ I I I ’ W S out to angles where it was also possible to obtain photographic recording. Actually, the counter apparatus shown permitted reliable measurements to be made over about twice the angular range covered by the photographic method. hforeover the time required to collect the data with the Geiger counter was only about onequarter the time required for the equivalent photographic esposure. 10
SPECIMEN-SLIT DISTANCE 300 MM. LINES - PHOTOGRAPHIC POINTS - GEIGER-MUELLER 10
t
t z c w
0
0
ASRECEIVEO 1470.F.-48 HRS.
0- 1800.F- 8 HRS.
z
01 AS RECEIVE0
1470*E-48HRS.
0 01
I
20
I
1
I
I
40 60 80 130 IDISTANCE FROM BEAM CENTER IN M M I
FIG.7. Coniparison of Geiger-Mueller and photographic measurement of small angle scatter of heated gaimma-alumina.
Small angle scatter measurements using plane crystal monochromatized radiation and photographic recording were carried out by Shull and Roess (19) and using hent crystal focused monochromatic radiation and photographic recording by Griinier (2, 13) in the apparatus described earlier. Small angle scatter measurements can also be made with the Philips’
268
M. H. JELLINEK AND I. FANKUCHEN
x-ray spectrometer, only a few simple changes being necessary. Fig. 8 shows this appnratus. All that is iiccdcd is ~ t l icxtrn dcfining slit in tfic x-ray beam path placed near the axis of the instrument and a niodified specimen holder. The specimen is first placed between the x-ray tube and the ntltlcd slit and the small angle air scatter around the main heam is measured with the beam weakened by the absorption of the specimen. The specimen is then moved to its location over the rotation asis of the instrument and the scatter is again measured. After subtracting the blank from the latter meusuremeri t, the resulting figures represent the small angle scatter of the specimen.
FIG.8. X-ray sprct,rometcr wit,h nrrow indicating position of slit to convert for small angle scatter.
Neither the film methods nor the Geiger counter method, as described above permit the measurement of the scattered radiation in the close vicinity of the main beam because of its high intensit,yandits halatingeffect on the film. The use of absorbers is a tedious task even with monochromatic radiat.ion and film, and becomes hopeless when white radiation is
~ l P P L I C h T I O NOF X-RAY DIFFHACTION
269
present. In many cases, however, i t is rather important to know the scatter at very small angles, in fact right to zero angle, in order t o make as complete an analysis as possible. ‘l’he reason for this need is seen from Table I which presents the angle a t which the scattering has fallen to one-half its theoretical value a t zero angle for various particle sizes. It is quite obvious that larger part,irles require measurements close to zero angle and that for samples cont,aining a considcruble percentage of large material a significant part of scatter may be lost if the scattering at very small angles is ignored. ~-
T.4RLE I
--
- _ _ . _ _ _ - ~ _ _
~.
~
Iladius in A.
____
Angle 20 in minutes a t which intensity of small m g l r smttering has fnlloii t o Iialf of vnluc a t zero angle
63.0
25 50
31.5 15 8 200 7.0 400 3.9 Computed for copprr h’a radiation __ --~ ____. __ - _ _ - _ _ _ _ _ _ ~ 100
It appears that this difficiilty may be overcome by a different method (Fsnkuchen and Jellinek, 20; Tarren, 21) of measuring small angle scattering (“Tn-o crystal method”). Fig. 9 is a schematic drawing of such an FIXED
SPECIMEN
L
ANALYZING CRYSTAL
L GEIGER TUBE
FIG.9. Scheinutic represetitation of the two-crystal small angle apparatus showing the defining slit, fixed crystal, spccimeii position, niialyzing crystal, Geiger tube, and optical pnth.
apparatus. A calcite crystal is set up as a monochromator, which because of its perfection reflects a beam of very narrow width. At a reasonable distance from this fixed monochromatizing crystal, another calcite crystal
270
M. 11, JSCL1,lNEK A N D I . F.\NKUUIIEN
is set up 80 that it can reflect the beam from t.he first crystal. Thus, when reflections from both crystals occur, the two reflecting crystal faces are parallel. If this sccond "analyzing" cryslid is rotated J m u t an asis lying in the face of the cryst.al at, the point where the reflected beam from t.he first cryst.al hits it, the beam leaving the second crystal and falling on the Geiger tube will vary in int,cnsit.y. In this manner, the angiilar dependcncc of the intensity from t,he first crystal can be dctermined. A sample can !400
250
-
200
-
150
-
fn
2000
1600
> k
W
5
100
50
1200
-
fn
+ z_
eoo
0
6
4
2
0
',"(",'
7.14'
18'
2
4 I
22'
26'
400
SCALE READING
FIG.10. l'\wwystrrl 2nd Iiod for nieasuring thc sitiall angle
scrtl ler
of gamma-
alumina.
then be inserted between the crystals and the sc:attel*ingagain me:isured. The difference between thc scattering before tind after iriscrtion of the sample represents small anglc scatter of the specimen through to zero angle. A correction for air scatter must also bc applied to these measurements. Because of the weakness of the doubly reflected beam, only the
APPLICATION OF X-RAY DIFFRACTION
27 1
scattering a t very small angles can be measured in this way. Fig, 10 shows the results of such measurements on a sample of alumina gel and Fig. 11 shows a complete small angle scatter curve made by combining t,he results of the standard and of thc two crystal methods. It is clear t,hat in this particular instance, neglecting the very small angle scattering would have meant the loss of the information on a very large portion of the sample. On very finely divided materials, however, results indicate that a study of the neglected region would not be very important. Ioooi The other straightforward method of measuring small angle scatter is that of Guinier (2, 13), in which the focused radiation from the bent crystal is passed through the specimen. If the camera is so designed that the specimen and focus lie on the circumference of the same circle, focusing conditions also prevail for REGION OF OVERLAP the small angle scatter and high 10 intensities are obtained suitable for photographic or Geiger recording. Fig. 3 shows this arrangement. DIRECT The two crystal mcthod is actually a procedure for using the double crystal spectrometer of x-ray spec0 5 10 IS 20 troscopy for the measurement of (DISTANCE FROM BEAM CENTER IN MM.p small angle scattering. Recently FIG.1 1 . Small angle scatter of alphaWarren (21) has described an ingen- alumina; coiribiriatiori of direct and two ious way of using the two crystal crystal methods of measuremcnt. spectrometer t,o measure what may be called total or integrated m a l l angle scattering. The crystals are both fixed a t the Bragg angles, i.e., to give maximum reflcction of the characteristic radiation. Three readings are then taken. The first is made with no specimen in thc beam, t.he second with the specimen between the two crystals and the third with the specimen in front of the Geiger counter slit (which can be quite wide). When the specimen is between the crystals, i t subtracts energy from the incident beam due both to absorptive processes and to srnall angle scattering. The reading with the specimen in the third position will be higher because now the small angle
liz\
272
M. €1. JELTJNEK AND 1. F.ANKIJCHEN
scattering can enter the slit and be recorded by the Geiger counter. ‘L’hr! difference betwccn the last two readings measures the total small angle scattcring which can be placed on an absolute basis by using thc first reading. WhiIc this method does riot give the angular variation of scattering, it is, nevertheless, very simple and rapid. It would appear that a stanclartlized “Wtwren” number may prove to be a useful wiy of charactcrizing many finely dividcd matmialx.
1V. LATTICEC’ONSTANTB Lnt.t,ice corislitnt+sand variutions of t.hese constants have occasionally been discussed as being rclatctl to the specific activit’icsof cert,ain c:it.Jysts. The poor x-ray diagrams usually obtained with high snrface materials do not lend thcrnselves to such determinut’ions,particularly if it is (:onsidered that t’he Iatt’icc coristant.s in the uppermost, surfncc layers of the catalyst \vould have t.o be measured in order to give information on the catdytic propert.ies. The dimensions of a. latt,ice of a given crystal are usually a function of t,he temperzhire nt. which the crystal is held. However, if other elements or compounds can go into solid solut.ion in this crystal, changes in the lattice may occiir. There are three major types of solid solution: 1. Suhstitutional solid solutions are characterized by a replacement of the atoms of subst.ance A hy those of H in all proport,ions for materials forming cont.inuous solid solutions or in limited tmounts for t,hose giving limited solid solutions. This replncement of substance A hy B may be complctcly random arid is designated “disorder.” 13y careful annealing, tlicsc disordcrcd rnnterials m:tv sometimes lx! converted t o an orclcid st,ruchtrirc where each tvpe of itt,orn occupies u specific position in the st’ructure. This latter effect forms a “superlattice.” Solut.ions of this t,ype arc regarded ns favorable possibilities if the radii of the atoms replacing each othcr do not diflcr by more t,han 15%. 2. Interstit,ial solut’ions itre characterized by atoms of smaller atomic size fitting into the int,erstitres bct.ween the largcr atoms of the structure of nnot,her solid. The iron-c:trbon system is an esamplc; t,here thc carbon atoms fill the spaces bet\t-een the iron atoms. 3. Siibtiwt,ive solut.ions itix? formed w h n the solute appears to extend the solvent, so t.hat not only does replacement take place but defect’sappear due to the abscncc of a cert’ain proportion of t,he solvent atoms from their expected posit.ions. X-ray diffraction providcs a vary sensitive tool for studying thesc effects, provided the changes conccrn the entire solid phase, and not merely one or a few atomic surface layers. This latter phenomenon, sometimes assumed for the esplanation of catalytic activities, is not detectable
APPLICATION OF X-RAY DIFFRACTION
273
by the x-ray methods so far developed. Although the change in lattice dimension is usually relatively small, being less than 1% in most cases, the form of Hragg's law allows very accurate measurements to be made a t certain angles. If Bragg's law is differentiated and the terms combined
ae
the result is -- = - 1tan 0. In the vicinity of 0=90", small changes in the ad d interplanar spacing d , due to lattice changes, cause large changes in 0. As e is measured on the film, highly accurate measurements of d , and consequently the lattice parameter, are possible if the reflections can be measured at, sufticiently great angles.
PART 11. EXAMPLES O F APPLICATIONS
1. IDENTIFICATION A few choscii illustrat,ions of the utility of identification studies in certain fields of catalysis will be discussed in some detail. Various mixtures of barium carbonate and ferric oxide (Erchak and Ward, 22; Erchak, Fankuchen, and Ward, 23) mere mechanically mixed and mere subjected to heat treatment at temperatures ranging betmee n 500 and 1000° C. The problem was then twofold; namely, t o identify the reaction products and to determine the catalytic activity of the products, and if possible to relate such activity to the reaction products. As all the compounds contained iron, filtered iron radiation was used to minimize fluorescence, and thin pyrex fibers coated with the specimen were used for the photographic recording of the x-ray diagrams while specimens mixed with vaseline on microscope slides were used for all measurements on the x-ray Geiger counter spectromcter. All diagrams were measured as to the relative intensities of the lines. One unique line of each compound, not interfered with by lines of any other compound formed, was carefully measured for intensity determination after careful scanning, by counts, on the spectrornetcr. Thus, if intensity of the chosen line was plotted for a given temperature as a function of composition, an extropolation to zero intensity would give the composition a t which that phase disappeared. Over one hundred preparations of the two initial constituents were made and studied. Fig. 12 shows typical powder diagrams obtained in this study along with the compounds identified. The changes occurring are clearly indicated on these diagrams. Similarly, curves wcre plottcd of the intensity of the chosen line as a function of temperature for a given composition. Obviously conclusions as to thermal stability, etc., can be arrived at from such curves. To follow the changes occurring in these experiments, the use of contour
274
M. H. JELJ,INEK AND I. F.QNBUC!IIEN
diagrams was introduced. I n these, a graph was constmcted with t,emperatwe as t!he ordinate and composition as the abscissa. Then, for each phase, t'hc int>ensityof t.he (oliosen line for each pair of tempernt,ure-composition
FIG.12. S-ray diffraction patterns of products obtuiricd at various tcmperatures from bariuiii cnrbonatc-fcrric oxide mixtures contailiillg i O % , ferric oxide: A, ferric oxide; B, barium carbonate; C, barium mi&; D, l3rrO.2FctO3; E, naO.GFerOs; F,n:ixF~~,O,,; H, extra lines in pattern of F.
APPLICATION OF X-RAY DIFFRACTION
275
values was plotted and points of equal intensity were connected. Such a contour diagram is shown in Fig. 13. This diagram shoxvs that most barium oxide was present in the samples which initially contttined between 35 and
FIG.13. Varistiori i n conceiitration of biiriuni oxide with compositiori "<mixture and temperature of preparation.
55% BaC03 and after heat treatments of the samples at temperatures above 725" C. It was even possible by this method to locate completely closed regions indicating the foimat2ionof compounds of limited thermal stability. Fig. 14 shows such a situation for the compound Ba0.2 FeaOr. In order to complete the identification of compounds formed in this reaction, Fig. 15, giving the regions of formation of compounds E and F, is shown. These compounds have been tentatively identified as Ba0.6 Fe203 and Ba8Fe&, by annlogy with other known compounds. Thus by the use of the contour diagrams, the compositions of all the specimens were determined tmd this justified a n attempt to correlate the catalytic activities of these same mixtures with their constitution. It is frequently the case that substances prepared by the interaction of solids in the absence of fluxing materials are in a finely divided state, provided that, the temperature of preparation is well below the melting points
27(i
M. H. JELLINEK AND I. FANKUCHEN
300
700
500
..
100%
Fl!&1
Li'cight per ccnt BaC03
100-3BaCOs
F I ~ :15. . \?urint.iori iri coiiccntrat,ion o f BuO-GFr201 (I?) nnd haFesOa, (F) wit.h composition of niixturc nrid teniprraturc nf I)rcpnrrit i o i i .
APPLICATION O F X-RAY DIFFRACTION
277
of any of tho reactants, or products, or their misturcs. In the finely divided form, siic*li nintc4als arc :is L: rule in a very suitable physical condition to cxhibit catalytic: properties. 'rhc pnrticu1:w iwiction stridictl for these catalysts was the catalysis of thc oxidation of carbon monoside. Typical timc-conversion ciirves for this reaction nre shown in Fig. 16 for three of
I rInp"ratur(!, "C!.
r 1
FIG.16. Catnlytic activity or wver:tl catidysts ns n function of teniperalurc. in t h e rcactian CO+ 1/20,-.COl.
the catalysts prepared at 600" C'. and it. is clear from this diagram that the catalyst made from 70% UaCOa is more active than t,hat made with either 10 or 40%, because higher tcmperatures w r e nccessary to attain the same conversions with the latter t,wo as w i t h t'he first ment,ioned. For piirposcs of direct comparison, t h c * t~empcrnturea t which the cntalyst gave 10% conwrsion was chosen as the reference point because the slopes of the curves a t this point were most' fitvorable. The 90yo point. gave very similar results with a few minor escept.ions. For t,he poorer catalysts, this ternperature~~asnbo~it 240"C., while the most, active catalysts gave t,his conversion at -40" C. Numbers of from 1 to 15 were assigned to cover this range in 10" steps. 'lkis, 1 represented 240", 2, 220", et,c. Jn this way, the more active cat,alysts were assigned the higher numbers. T o corre1nt.e t.he cat.alytic propcrtics of t,he samples wit,h their preparat,ive temperatures and also with t.he results of thc x-ray analysis, it \ m s found rathcr convenient to consf ruct mother contour diagram. The result of t.his construct.ion is shown in Fig. 17. This diagram has the composition of t.lie shrting mistuw iiwd in prcytiring the catalyst. :LS :iIwissa rind t.hp
278
Af. H. JELLINER A N D I. FANKITCHEN
temperntiire of t.he lieat t,reatrncnt cmployetl for the catalyst prcpwnt ion :is t.hc ordinale. 'I'hc numbers representing the 1tctivit.y of 1.he individual samples ~vei'eplaced a t the appropriate positions and then all similar numbcrs were connccted in such n. manner that, 110 lincs interscctcrl. Inspect'ion of this figure rcveals t,hat, there cxist cert'ain rcgions of maximiim act,ivity.
In the orclcr of decreasing activity, they m e 70.-850, 70-700, 90-700, 50-750, ctc., and there arc regions where the activit'y falls to n, minimum; for example, 30450, 45-600, 70-500, etx. The first, number refers to the percentage of UaCCls in the i n i M niixt.iire, the second number to the t!emperature i n degrees C. to which the mixture was heat,ed. The a c t k and inact'ive compositions could be looked up on the x-my contour diagrams shown earlier but a far more convenient method consists in merely plot,t,ing the x-ray determined con tours itrid the act.ivit*ycontours t-u t.he same scale on transparent paper and siiperimposing them to determine t'he uc tive phases. As an example of thc results obt'ainahle, the cat.itlysts obtained from starting mixturcs with 707' I3aCOa will be discussed in more detail. The rapid rise in activity of thcsc c:tt.nlyst.s t o 14 for ciitalyst preparation temperatures behveen 500 anti 700" C. is followed by a decrease to 11
APPLICATION OF X-RAY DIFFRACTION
270
for a catalyst prepared at 800' C. and by a further increase to 15 for a catalyst made at 850" C. and a subsequent decrease to 8 for the catalyst made a t 900" C. These catalysts were prepared from mixtures which gave the maximum amount of formation of BasFesOpl. No evidence of the existence of this compound was found below 650" C. A possible conclusion to draw from these observations is that the observed activity was due to this compound in small quantities or in a poorly crystallized state. As more is formed, an increase in actiyity is observed. Some of the variations may be attributed to the fact t h t this initial composition is just at, the point of maximum formation of UasFes021and, in a region of rapid change, small inconsistcricies mav 1)c cspectetl. In this manner, all the various portions of the constitution diagram can he, at, leiist tentatively, interpreted and rc4:itetl to thc cat:ilgtic. propert ics of the solid phase for the ('0 osid L: t'1011.
FIG. 18. Debye-Sclierrcr diclgr:mis of iron Fisclicr-Tropsch catcllysts; reduced I'iedi ciitdyst, used ccltdyst, used catalyst, pure iron, pure iiiagnctitc.
The Fischer-Tropsch Synthesis, i.e., the catalytic reduction df carbon Inorloside by hydrogcn t o liquid, gnseous, and solid hydrocarbons itrid oxygenated organic compounds, has recently bccn reported to operate efficiently under superatmospheric pressure, ivith catalysts which con tairi iron as their basic component. More recently catalysts of this type are being used in B very fine dispersion which permits one t o operate with them in a "fluid bed." Such a catalyst usually consists, before the start of the synthesis reac-
280
hl. 11. JICIzLINKK .4ND 1. F A N I U C F I P N
t,ion, o f magiieLiI.e, FesOr,lvith a very large rrystdlite size. This large sizc of t.he crystallit.es is evidenced by the very spot.ty character of the powder diffraction diagr:ims. If the catalyst is reduced at temperatures of, say, 1500" F. in hydrogen, thc product yields a diagram of metallic iron in a siiffiricnt,ly finely divitled foim to yield srnooth diffract'ion rings. Such a pnttei-n is shown in the uppermost. diagram of Fig. 18. -4sa comparison, a standard pattern of iron is shown as t'he nest to t'he bottom patt'ern. The prccisc agreement in the p:tt't,c?rn of t'hc lines is unrnistakable. Aft,er some use of tlic c:italyst in t.he synthrsis reaction, a sample can he wit~lidrawn while thr catalyst is still :it, :i high level of activity nnd an x-ray diagram is made of this sample. (See t>licsecond tliaglntn of Fig. 18.) It is obvious that t,lic met,allic iron has nearly :dl vanished and hus becn replaced almost eritirely hy rn:ignetitc. .4sa compirison, the f i r i d diagram ( ~ Fig. f 18 shows the stand:irtl pattern of magr1clit.c. A smipie IKLS ulsoobtainedof :L rat.alyst, \vhich hat1 been operated for a longer time but wi.s still a.t a Iiigher level of :ictivity. It,s diagram is t.he niidrlle one in Fig. 18 : ~ n ds h o w that the magrid ite has been convcttrtl by a p h s e change to some other compound. N'hile the at.andard pat.trrn of this material is not' inctlutletl in the figlire, it's irlen t.it.y has hccn c.stablishetl as a low tempernturr carbide of iron, probublg Fc?('. 011 Iicxting, t,liis compound can be changed to FeC, cementite. The c-orit,initoilsactivity of thc r.at.nlyst'through t'liesc changes in structure is puzzling. 1'crli:ips onc shniild not, speak of one single " Fischer-Tropsch c:it:tlyst."'! Xtcr somc: operation in the synthesis r e a h o n , it, ocrcnsionally becomes dc.sir:tl)le to treat the c:it,alyst,:kt :in elevated t.emper:tturc 1vit.hhydtwgen in order t o remove some of t,he ~ r a s ycompounds adhering tau it. Ikforc the hydrogen trcat,ment, (provided 110 high prcssurc riins wilh the cat'dyst had I)cen made), t.he cnt,nlyst is mainly mugnet'ite, plus :t little iron, : i d perhaps a t r w c of cnrlide. X-rny studies shon- thut 1he hydrogen txeatrncnt converts the rntalyst back t'o iron. 111 :L few :tpparentIy complet.ely random Fisr:her-l'ropsch catalyst.s, rstt,hc.r large>tlist,ortioris of the iron and magnet.it.e1at.t.iceswere noted. This may be (lite t,o solid soliit.iun o f a 1wiet.y of tmipunents.
11. C'RYSTALLITES
AND
I".~KTICLRG
I3ecaiisc of the close connection 1)ct ween these two phases of x-ray study, especiully in the work discussed here, they will he considered together. (Riscoe arid \\'arreri, 24; ,Jellinck and Fnnkuchen, 5 ; Rows and Shiill, 19.) Artivatccl alumina purchased from the Aluminum Ore Company was given ;i treatment with dilute hydrochloric acid to lower the sodium oxide
APPLICATION OF X-RAY DIFFRACTION
28 1
content from about 0.57, t.o less than 0.1%). This alumina, designated "low soda alumina," \\-as then given a series of heat treatments of increasing severity. The first studies on this alumina were of :I simple exploratory nature to determine what changes in diffract'ion effects, if any, could he detected with changes in catalyst activity. .4part of t.his study \\as carried out on the alumina base. In further studies, it developed that the presence of molybdenum trioxide acted as an acceleritt'or to the changes t,hat occurred in t.he pure alumina. I n other words, addition of the molybdena brought about a certain change in t'he alumina at' ti t.emperature which was 100' C. or more lower t.han the temperature at which a simi1a.r change took place in t'he alumina base. Fig. 19 shows t.he powder diagrams of the lon- soda alumiris bhat had been given the indicated heat treatments. The original sample received from the Aluminum Ore Company gave t,hc diagram sho\\n tit the top of this figure and was easily identified as gamnia-nliimina. As this material was heated to higher and higher temperatures, not only did the lines in the diagrams sharpen, as can clearly be seen, but. the number of diffraction rings increased. For several reasons, this seemed t'o indicate riot the formation of any new form of aluminn, but rather a development and perfection of the original structure. I n the more highly heated samples, the presence of corundum or alpha-alumina can lic delcctcd by comparing thecorresponding diagrams with the final pattern! which is that' of alpha-alumina. That changes in particle size also oceurrcd wis indicat.ed by the phot,ographic small angle scatter diagrams (Fig. 5 ) . These basic changes of gro\\th of cryst'dlites and pnrtkles, and of development, of the st,riictiire of t.he aliimina h e can he shown to have rather important correlations n-ith the siwfnce areas of the catalysts and their act,ivities in hydroforming. One of t'he rather import'ant technical problems concerning these cat'aIysts \\-as the prolonging of their useful life. It was found that high temperature heat treatment of hydroforming catalysts represents an accelerated test by result.ing in the same changes as occur a t much lo\wr t'cmperat'urcs over cxtcmkd pmiods of t'ime in acbual field service. 'I'herefore, in studying new base materials for st'ability, it was a cust.omary procedure to heat t.hem to a given t.est t.emperature, check their activity, heat treat, t'hem more severely and continue until a temperature was attained which caused a drop in activity. This was a costly and time-consuming process. After the advent of t'he x-ray techniques, a shorter test method was developed involving heat treatments of small snniples, followed, instead of the usual activity test, by an examination of t'he powder x-ray diagrams. This permits an easy and fast detection of any dct.crioration of the catalyst properties by overheating becaiise of t'he very marked change in the x-ray
282
M. H. JELLTNEK AND I. FANKUCHEN
1
11
Fro. 19. Dcbye-Scherrer diagrams showing thc treatment of low soda gamma-alumina.
etfect of increasingly
scvcre heat
APPLICATION OF X-RAY DIFFRACTION
1200" F. 1 hr.
1470" F. 1 hr.
1470" F. 353 hr.
1560" F. 24.5 hr.
15tiO" F.
2-10 hr.
1ti50' 1:. 1 9 hr.
1650" F. 282 hr.
1740" F. 2 hr.
1740" F. 6 hr.
FIG.20. Dcbyc-Scherrer disgrttriis of heated alumina gel.
283
M. H. JELLINEK AND I. FANKUCHEN
284
patterns. If the catalysts themselves had behaved as did the alumina arid had shown gradual changes, the task might have been more difficult and
1785" F. 2 hr.
1785" V. 12
1830"
h.
v.
2 hr.
1830" E'. G hr.
1880" I:.
ti hr.
1880" F. 2 1ir.
Only 1880" F. 2 lu.
1920" F. 5 hr.
FIG.21. Debye-Schcrrcr diagrams of hcltted aluminn. gel.
APPLICITION O F X-HAY DIFFRACTION
285
t,imc-consuming, but the presence of the molybdena seemed to make the change of the s-ray p:ittcrns :in d l or not,hing effect. The nced for obtaining a more stable alumina ctttalyst led to the study of gel aluminas and it is with some of t,hese gels that the next, discussion will be concerncd. The part'icular :tlumina gel to be considered here was prepared from aluminum nitrate which had been precipit'atcd by ammonia and washed. I t was then subjcctcd to a few heating tests which quickly revealed that it was far more st.ahlc t.han the low soda alumina. Thereafter, n sample was plt~cctlin an clect'ric muffle and small samples rcmovcd pcriodically. \\'hen little or no fiirther change in the powder diagrams or the small angle scattcr patterns could be noted, t,he temperature was raised arid the heating continued. 'l'hus, each sample had a ciimulative heating. Figures 20 and 21 show the powder diagrams of somc of t.he samples, as well as the t:emperatiire at, whicli thcy\\-cre heated. The st,arting sample \\-as nearly amorphous. The heating at. 1470" F. for 353 hours produced a distinct change as :t comparison of X-4 and X-12 show. Aft'er t>his,n much slower change talws place hecniisc X-13 is not fnr different from X-12 and X-20 just shows ii sligh t, ati:itpenirig and development of the lines. Raising the temperature to 1650" F. shnrpencd t.he diagram consider:tbly but furt.her heating at this t.emperature \\-:isof smdl ctrcrt. The remaining cliitgranis show the resiilt s of heating sninll smiples of essentially S-30,at. t'emperatures \myiitg from 1740 to 1920" I?. A most, int.rrcsting p:Lir itmorig these is X-47 and X-48. l'hcy \\-crc both heatccl a t 1880" F. for 2 hours but, the former had had a prelimina.ry cwnulat.i\-e hcat.ing while t.he latter had had no high t.emperctture treatment prior to the final heating. The cffectt.~of the eai*licrhcntings on X-47 are distinctly wen i n the sharpness of the lines as contrastcd to those of X-18. r . 1 he awrnge c.ryst:illite sizes of certain of these specimens whic*hs l i o ~ c d n o detectable trace of corundum were detcrrnined from s-ray spec.troniet.er runs on the 1.40 A. linc. The crystallite radii run from 42 to 79 .I., the sizes increasing with increasing lieat'irig temperatures. There appear t,o bc some slight iriconsistencies but, t.hey are within thc limit,s of t,he esperiment.al errors. 'I'he trcnd is distinct,. Here, too, it is of interest to (.omp:tre X-47 and 5-48. The lntt'er has only increased its crystallite size hy 50Oj,, while the cryst.al1it.wof t.he formcr have just about doubled in size. Small anglc x-ray scattering st,udies were also made 011 these catalysts and the usual mcasriremcnt,s nrid imalyscs of the data were m:rde. The rcsults w r e plotted with the particlc sizc as abscissa and t'lie weight per cent, of t.hc particles of t,his particle size as ordinate and t.he datum p0int.s were connected by a line to yicld the simulated distribution curve. (Jellinck, Solomon, and Fankuchen, 18). Tt is obvious that a widc range of particle
286
M . II. JELLINBK A N D I. FANKUCIIEN
sizes is present, sizes right up to the upper limit that can lw reached by the straightforward mcthocl of small angle scatter study. ‘I’his suggested that studies made t’hisway \wulil cover o~ilyt,lie part of t.hc specimen composed of pnrtkks below it ccrt,ain limit of’ size. ‘I’he two crystal method oUers the possibility of raising t’hislimit. It, is useful to determine which part, of a given spccinien responds t,o sm:tll angle meitsurements, i.e., whether ccrt.:tin part.s of the specimen consist cithcr of particles t,oo large or too small to be studied by this mct hod. From the theoretical discussion in bhc. first part of t,his pitper, it can be shown that the int’ercept K’Yof successive tmgciits to the scattering ciirve (Jellinek, Solomon, and Fankuchen, 18) clivitled hy Ti3 (obtained from the slope) give quotient,s which would be proportional 1.0 the weight of the particles of this specific size. If esperimentnl condit.ions arc rigidly cont.rolled so t’liat.t’he same w i g h t of specimen is in t’he x-ray heam in cach experiment, and if, for each s:trnple analyzed, t.hc Z,,K,,.’Rn3is evaluat.ed, the value of t,his sum, “111” is proport,ional t o the amount of sample contributing a nieusurablc scattering. If one of the mildly heated samples consisting alrnost. esuclusi\dy of very smtill particles is arbitrarily selected as a base and it is assumed t.hat all its particles have measurable scattering, all ot.her samplcs ran be related to this one. Thus, thc “R‘l” value of each sample is divided by t.lie “RZ” value of t.he st,andard an11all the weights of t,hc fractions of varying size pnrLicles of t,he sttinple under cmsideration tire mult.iplied by t.his fraction. This has the effert of correct.ing t’lie wight, dist.ribut.ion by a factor deterniincd by t.he fr:wt.ion of tlie sample that, consists of particles that have grown too h r g e to be measured by this method. The results of such small angle analyses on certain dumina gel samples is illusti~nt,edin Fig. 22. The solid point,s are t,he actual values of radii and weights rletcrmined by analysis of small angle sc,ztt.er. 111 t’heupper right corner of each graph is cnt.rrd t.he value of “RI” lor that, specimen. The brol;cn lines arc thc: distributions corrected its just described and the arrows are a t the points where the average cryst.nIlite sizes \\,ere determined to be. Therefore, the shn.pcs of t.hesc curves give t.he dist,ribution of those particles which could I)e measured I)y small u17glr scatter m d the position of the broken line between t,he base line ard the solid line is an indication of the relative proportion of rnat.crinl t.hat. \\.as not measurable. These curves showed that the distrihiitions chiingetl in the sense of indicat’ing the presence of more larger purticics and also of more unmeasiirable material (because the particle size is too large) as the sevcrit.y of t’heheat treatment was increased. The fact that highly heated, or well crystallized specimens of alumina are not suitable samples for small angle scatter measurements
287
h PPLICATION OF X-RAY DIFFRACTION
with the standard apparat,us is shown by the distribution diagram of sample X-43,where the distribution of the measrired part iclcs representctl merely about 9% of the total sample. PARTICLE RADIUS IN
10
20
30
40
10
20
30
40
50
i
60
50 60 PARTICLE RADIUS IN
70
80
90
100
70
80
90
00
d
FIG.22. Thc wciglit distributions of particle sizes obtairietl from small angle scatter from hented rtluiiiiiia gcl.
Onc apparent contradiction milst be pointed out. Fig. 22 shows the results of some small angle studies on alumina gels. The arrows indicate the crystallite size as determined from line width measiirernents. In every case this arrow is a t n value of R greater than that corresponding to thepeak as determined by the small angle scattering. I t mould appear that by definition the particle size should always be equal to or larger than the crystallite size. There are several possible reasons for this discrepancy and it is probable
288
hl. 11. JF:LLINEK AND I . E'ANKIJI'HEN
that all of them are partially responsible. In the first place, t'he small angle nieasuremexits favor the small part,icles despite the correct,ion fact'or int'roducctl because this correction does riot, change t'he shape of the dist,ribut.ion curve. Secondly, t,he spcct,rometer measurement.s wliich were used for the crystaI1it.e size comput.ation were made continuously without taking individiial counts across the diffraction line, and flat, instead of curved, specimens were used. Both of these lust two effects cause the linc shapes t o become somewhat warped and thus introduce errors. Lastly, the very form of the line shape when there is a dist.ribut,ion of crystullite sizes murkedly favors large crystallit.es. The first, and last reasoris are probably t.he decisive ones, as they both tend to produce the observed disagreement between the results of both methods. Silica gel cracking catalysts have also been studied. These materials are amorphous and yield no powder diagrams but they do give a very marked small angle sc,atter. If the particle dist
PART 111. REFERENCES 1. Beeck, O . , Wheelrr, A,, and Smith, A. E., Phya. Hcz'. 66, GO1 (l!1391. 2. Guinier, A,, Ann. ph!ln. 12, 162 (1!)39). 3. Dcbyc, P., Physik. Z.28, 135 (1927;31,348(1930). 4 . Bragg, \V. L., The Crystalline State. Bell and Sons, Lundon, 1!133, 1). 35. 5. Jcllinek, hi. H., arid E'ankuchcn, I., Znd. Eng. Chert!.37, 158 (1945). 6. Fankuchcn, I., Naturr 139, 193 (1!)37). 7. I,onsdale, K., Pror. Roy. Sor. Lonrlua A177, 276 (1941).
j\PPLICATION O F X-RAY DIFFRACTION
289
Lipson, H., Nclson, J. B., nnd Iiilty, D. P., J Sr+. Instrirmmts 22, 184 (194,5). Johannson, T., 2. Phyzik 82, 507 (1933). Bozorth, R. hI. arid liaworth, E’. I<:.,I’hys. Rec. 63,538 (1938). DuRIond, J. LV. and Kirkpihick, H. A., Re?,.SCL.In..drictuents 1, 88 (1030). Smith, C. S., Rev. Sln‘. Instruments 12, 312 (1941). Guinier, A., J . p h i t n . phys. 40, 133 (1943). Hosemann, R., Z.Physik 113, 751 (1939). ICratky, O., Sckorn, A., and Treer, R., 2.Elektrchem. 48,587 (1942). 16. Bauer, S. H., J . Chem. Phys. 13,450 (1045). 17. Roess, I,. C., J . Chem. Phys. 14, 695 (1946). 18. Jellinek, RI. H., Solomon, E., and E’nnkuchrn, I., l n d . Eng. Chem., . I d . Ed. 18, 8. 9. 10. 11. 12. 13. 14. 15.
172 (1946). 19. 20. 21. 22. 23. 24.
Roess, L. C., and Stiull, C. Q., J . Applied Phys. 18,295 (1047). Fanliuchen, I., aiitl ,Jcllinek, hl. H., Phys. Rev. 67,201 (1945). Warrcn, B. E., I’npcr preswitcd at the A.S.S.R.E.D. Rlerting, June, 194i. Erchak, RI., and Ward, R . , J . .Ant. Chew. SOC.68,2093 (1946). Erchnk, RI., Fankuchen, I., and Ward, It., J . A m . Chent. Sor. 68,20&5 (1946). Biscoc, J., and IVarrcn, 1%.E., J . Applied Phys. 13, 364 (1942).
This Page Intentionally Left Blank
Author Index Names in parenthesw indicata senior authors of the references and are included to m i a t in locating referencee where a particular name is not on a given page. Elample: Anderson, J.. 33 (see McAUistar) menu8 that McAlliater d d.will be mentioned on paKe 33 the cl J.accounting for Anderson. This article can be 1ocated.undey McAIlietar i n the list of mferencea: B which references are hated in bibhographies at the end of each Numbers in italics refer to the D. B-C ~ ~on article.
A
Abraham, B. M., 240, 243 (35),256 (35) Aicher, A., 141, 142, 166 (80) Alden, R. C., 47,63 (30) Aldrich, R. C., 119 (341,164 (34) Alekseevskii, E. V., 111, 114 (36) Anderson, J., 33 (see McAllieter), 38 (see McAllister) 42 (see McAllister), 56 (see McAUiter), 57 (see McAllister), 58 (see McAllister), 63 (12) Anderson, R. B., 79,85,89 (25,26), 90 (41),130 (55),141 (55),142 (55),150, 152 (90),166 (55),166 (90) Askey, P. G., 75, 89 (11) Atwell, H.V., 118 (42),119 (42),121 (42), 123 (42,50, 531,164 (42,50),166 (53)
B e m n , H. E., 133 (65), 166 (65) Benton, A. F.,13,26(1 l), 66,67,89 (2,6), 190, 191 (111,116),198 (110,111,116) Bevan, D.C.,191 (112),108 (112) Bilte, W., 103,113 (21) Birch, S.F.,32,38,56, 59,63 (10,El) Biecoe, J., 280, 289 (24) Bishop, J. W.,245,266 (44) Bleakney, W.,175 (64),176 (64),197 (64) Bloch, H.S.,218,266 (19) Blunck, F.H., 52,64 (35) Boedeker, E. R.,202, $66 (2) deBoer, J. H., 66,80 (3),163, 171,196 (la), 187 (41) Bonhoeffer, K.F., 173, 174 (57,sn), 180, 187 (53, 56, 57, 61, 74) Bonnell, W.S.,47, 48, 49, 64 (32) Bosworth, R. C. L., 159, 163, 164, 165, 170, 173, 183 (7), 196 (6,7, 17, 24) Bozorth, R. M., 261, 989 (10) Bragg, W.L.,260,262, 288 (4) Brice, G.,172, 174, 197 (49,50) British Intelligence Objectives Subcommittee, 100 (15),113 (15) Brooks, J. A., 229 (32),237 (32),238 (32), 240 (32)266 (32) Brown, C. G.,81, 89 (35) Browp, H.C., 31, 63 (6), 217, 266 (18) Brunauer, S.,2, 3, 4, 11, 21, $6 (3,5, lo), 66, 67, 68, 69, 70, 72, 75, 77-?9, 82, 83 (4,391, 84,89 (4a, b, 7, 12,39,40) BufIleb, H.,143, 146, 147, 166 (83,85) Burk, R.E.,91,113 (I), 244 (a), 245,252 (421,254 (421,266 (42,44) 1 Burstein, R., 175, 176 (67), 180 (76), 197 (67,76) Burwell, R.L., 3 (see Taylor), 22, 26 (6f)
B Badische Anilin u. Soda Fabrik, 115, 163 (1) Bahr, H.A., 138 (fl),166 (71) Balandin, A. A., 91, 113 (21,187, 189,I98 (99b,104, 105) Balanova, T.F., 104 (n), 113 (22) Ballard, 9. A., 33 (see McAllister), 38 (see McAllister), 42 (see McAlliatsr), 56 (see McAllister), 57 (see McAlliiter), 68 (see McAllistm), 63 (12) Bartlett, P. D., 34, 35, 39, 40, 42, 63 (151, 218,266 (22) Bauer, S.H.,264, 289 (16) Bauermeister, H.O.,88 (461,90 (46) Baxendale, J. H.,186, 198 (99a) Beatty, H. A., 252,254 (45),266 (45) Beckwith, J. B.,78, 89 (20) Beebe, R. A., 23, 26 (18),78,89 (20) Beek, O.,2,13,26 (4a,b), 94, 113 (Q),159 (9),165, 166, 167, 168, 187, 194, 196 C (9,33d),223, 268 (25a),259, 288 (1) Benedict, W. S., 13 (see Morikawa), .# Caesar, P.D.,33, 63 (11) (12a),181, 184 (80), 197 (80) Calingaert, G., 252,254 (45),266 (45)
292
AUTHOR INDEX
DuMond, J. W.,261, 889 (11) Campbell, D. H., 130 (56),166 (66) Dunsten, !4. E., 32,38 (weBirch), 56 (see Cantuniari, I. P., 209,866 (10) . Birch), 69 (see Birch), 63 (10,21) Carlson, G.J., 224, 244 (28), 268 (28) Carmody, D. R., 41,42,52,58,59,63(X), 84 (35) E Caasie, A. B. D., 79, 89 (21) Eckell, J., 111, If4 (38) Cavanagh, B.,193 (125),198 (125) Cawley, C. M.,108, 114 (30),117 (36), Egloff, G.,28, 63 (2), 218, 244 (21),868 (21) 119 (36) 164 (36) ChatTee, C. C., 118 (451,119 (451,123 (45, Eley, D. D., 24 ( N ) ,26 (20a, b), 169 (8), 165 (23a)166 (28),176, 177, 178,179, 53), 164 (45), 166 (53) 180 (73),181 (73),183 (M), 194 (28), Chapman, D. L., 191, 198 (113-115) 19B (8, 23a,28), 197 (73),198 (86) Cheney, H.A., 223, 266 (26) Chenicek, J. A,, 47, 48, 49, 83 (31),223, Elgin, J. C., 190,191 (lll),198 (110,111) Elkin, P.B., 80,89 (30) 224 (27),868 (27) Elvins, 0. C.,116,163(5) Christian, W.,192 (120),198 (120) Emmett, P. H.,2,2,4, 11, 21,88 (3,6,7, Ciapetta, F. G.,36, 37, 83 (18) lo), 66,67-70, 72, 75-82, 83 (39,4), Cines, M.,87 (45b),88,90 (46b) 84,87 (45b), 88,89 (q8, b, 7, 12, 17, Coehn, A., 171 (46),197 (46) 19, 26, 27, 28,33, 34, 39, 40) 90 (454 Condon, F. E., 34, 35, 39 40, 42,63 (16), b), 139, 162, 166 (76),180 (75), 191 218, 266 (22) (116),197 (75), 198 (116) Conn, G.K.T., 93,113(6a), 185,198(91) Erchak, M.,273, 889 (22,23) Constable, F. H.,188,198 (102,103) Eucken, A., 19, 20, 21,88 (15) Coulson, C. A., 196, 199 (135) Evans, M.G.,195 (133),189 (133) Crafts, J. M.,220, 256 (24) Evering, B. L., 226,228,231,238,240,242, Crawley, B., 109, 114 (32) 247,248,252,968(31,34,44a) Craxford, 8. R.,97,f13 (12),116,123 (61), 150, 163 (17),164 (51),168 (88,89), Ewing, W.W.,80, 89 (29) Eyring, If., 171, 173, (52),180 (52), 194, 187, 188, 198 (100) 196, 198 (37,29) 197 (52) Crowell, J. H.,133 (651,166 (65) Csech, H.,163, 196 (14) F D Fajans, E., 175 (63),197 (63) Fankuchen, I., 260,261,264,266,269,273, Davis, J. D., 110, 153 (7) 280, 285, 286, 888 (5, 6), 889 (18,20, Davis, R. T.,78, 84, (40),85, 89 (19),90 (40b), 166 (27),198 (27) 23) Faragher, W. F., 118 (45), 119 (46), 123 Debye, P., 259, 264, 288 (3) (46,62),136 (68),164 (45), 166 (52, Denham, H.,38, 83 (22) DeWitt, T.,78,79,80 (27),88,89 (19,27), 88) Farkee, A., 24 (21), 28 (21), 92, 94, 113 90 (45a),139 (751, 152, 166 (76) (68),174 (67,6l), 175 (65,66), 176, Diamond, H.,174 (60),197 (60) 179-181, 182 (83,84). 183, 184, 186, Dillhey, P.,138 (72),166 (72) 193,197 (67,61,65,66,69,71, 72,74, Dippel, C. F., 66,89 (3) 70, 81-84,) 198 (95, 126) d’Ouville, E. L.,226 (31), 228 (31), 231 (31),238 (31),240,242, 247 (31),248, Farkas, L. J., 24 (21),$8 (21), 92, 94,113 (5,8),174 (68),176 (a), 179,181,184 ,968 (31,34) (79),197 (58,65,72,79,81) Dowden, D. A., 23, 88 (18) Feachem, C. G. P., 76,89 (11) Drigan, A,, 244,254,868 (41) Dryer, C. G.,223 (see Chenicek), 224 (27), Fehrer, H., 103, 113 (19) Feisst, W.,117 (23),164 (23) 868 (27)
.
293
AUTHOR INDEX
Feitknecht, W., 111, 114 (37) Fidler, F.A.,38 (see Birch), 56 (see Birch), 59 (see Birch), 83 (21) Field, H. W.,57,62, 64 (41) Field Information Agency 101 (16),113 (16) Fischer, F., 116,117,124(21),138 (71,72), 140, 142, 144, 145, 148,149,163(24, 8, 19-22), 164 (23,25, 28, 29), 166 (71,72, 76, 77), 168 (82,84) Foran, E., 119 (341,164 (34) Fowler, R.H.,168 Francis, A. W.,33, 8.9 (11) Franck, J., 171 (47),197 (47) Franke, N. W.,202, 214, 966 (3) Frankenburg, W.G., 21, 84 (a), 85, 90 (@a), 165, 166,167, 168,198 (23) Frankenburger, W.,21, d8 (16),165, 166 (26), 196 (23b,26) Frazer, J. C.W.,110, 114 (35) Frey, F.E.,29,47(see Alden), 8.9(3,4,30) Frid, H.V., 111, 114 (36) Friedel, C., 220, 966 (24) Friedman, B.S.,55, 64 (38) Fujimurn, K.,116 (1%15), 141 (79), 263 (13-15),166 (79) Fuek, T. F., 80, 89 (31) G
Gall, D., 123 (511,164 (51) Gallsway, W. S., 61, 62,84 (43) Garner, W.E.,23,96 (19) GeckIer, R. D.,244, 268 (39) Gerrner, L. H.,164 (19), 196 (19) Giauque, W.F., 173, 197 (51) iGhbrook, A. L.,224, 968 (29) Glssgow, A. R.,Jr., 57, 62, 84 (40) Glasstone, S.,173 (521,180 (52),197 (52) Golden, P.L.,100,113 (14),116, 150 (16), 163 (16) Golumbic, N., 85 (see Anderson), 90 (41), 130 (55), 141 (56),142 (55),166 (55) Gorin, M.H.,33,63(13),208,209,956 (9) Gould, A. J., 175 (64),176 (64),197 (64) Gould, D.W.,57, 62, 84 (41) Greenhnlgh, R. K.,186,198 (97,98) Gregg, S. J., 77,89 (18) Griffith, R.H.,103, 109, 110,l l S ( 2 0 ) , 114 (32,W
Grosse, A. V., 27, 32, 37,39,43,44,54, 60, 61, 89 (8,23,25,26, 27), 84 (37),244, 245 (43),968 (43) Grumitt, O.,244,252,254(a), 966' (42) Guinier, A., 259, 281, 264, 267, 271, 888 (21,889 (13)
H Haensel, V., 118 (44),119 (441, 123 (441, 136 (69), 164 (44), 166 (69) Hall, C.C.,108,114(30),117 (35,361,118 (43, 44, 48, 49), 119 (35, 36, 43, 44, 48, 49), 123 (43,44, 48, 49, 51), 150 (871,164 (35,36, 43, 44, 48, 49, 51), 168 (87) Hall, W. K.,191 (113), 198 (113) Halsey, G.,21, 168, 198 (33~) Hark-, W.D.,76-79, 89 (13-15) Hahess, R. W.,4, 88 (7), 180 (75), 197 (76) Harteck, P., 174, 197 (66) Hartner-Seberich, R., 117 (24), 164 (24) Hawk, C.O.,86 (41),90(41),100,113(141, 116, 130 (55), 141 (55),142 (55),150 (la), 163 (11, 16), 166 (65) Haworth, F. E., 261, 989 (9) Heckel, H., 117 (34), 119 (34),164 (34) Heldmann, J. D.,53, 84 (36), 202, 203, 206, 215, 216, 242, 966 (1,5,6, 16) Hepp, H.J., 29, 47 (see Alden), 83 (3,30) Dr. Herbert, 125 Herington, E. F. G., 97,100,101,102(17), 105, 113 (13,17), 114 (24),188, 190, 194, 198 (101,109) 199 (130) Hilbcrath, F., 117, 118, 154 (39) Hill, T.L.,79,89 (22,23) Hinshelwood, C.N., 158,170 (3),193,194 (31,195 (3) Hirota, K., 171 (38),198 (38) Hixon, R. M.,108, 114 (26) Hodler, A.,21 (seeFrankenburg),98 (16b), 166 (26),196 (26) Hofer, L. J. E., 86 (see Anderson), 90 (41), 130 (55), 138 (73),141 (55), 142 (55), 150, 151 (91), 152 (90),166 (65, 73), 168 (90,91) Kogneas, T. R.,192 (121), 198 (121) Holloway, C., Jr., 47,48, 49, 6'4 (32) Honig, J. M.,78, 89 (20) Hoog, H., 203, 226, 239, $66 (4)
294
AUTHOR INDEX
Keith, P. C., 130 (59, 611, 155 (59, 61) Kernball, C., 13, 86 (13) K i g , J. ~ . , * i 2(531, 3 165 (53) Kingdon, K. H., 163, 196 (16) Kingman, F. E. T., 23, 26 (19) Kirkpatrick, H. A., 261, 289 (11) Kistiakowsky, G. B., 165 (21, 22), 198 (21, 22) Koch, H., 116, 117 (21, 25, 26, 28, 31), 118, 124 (21), 163 (10, 21), 164 (2426, ZS, 31, 38) Kodama, S., 116, 165 (12, 13) Kohlschutter, H. W., 22, 26 (17) Komarewsky, V. I., 27, 43 (see Ipatieff), 69 (26) 218, 244 (21), 856 (21) Kratky, O., 264, 889 (15) K h , C.S., Jr., 33, 63 (13) Kuhn, R., 192 (122), 198 (122) Kumari, Z. I., 108, 114 (29) Kurnmer, J. T., 139 (79, 152, 156 (75) Kuater, H., 140 (76), 166 (76) I Kvetin8ka8, B., 214, 239 (13), 240 (13), Ipstieff, V. N., 27, 32, 39, 43, 44, 54, 55, g66 (13) 63 (1, 8, 23, 25, 26, 27), 64 (37, 381, 214, 223, 226, 239 (13), 240, 243 (351, 244, 245 (43), 254, 255 (13), 266 (30, L 35, 43) Laidler, K. J., 173 (52), 180 (52), 197 (62) Iverson, J. O., 223 (see Chenicek), 224 de Lmge, J. J., 95, 119 (10) (27), 366 (27) Langmuir, I., 1, 86 (11, 69, 74, 89 (8, 9), 158, 163, 165, 166 (29), 171, 172, 193, J 196 (16, 23c, 29, 33b, 40) Jellinek, M. H., 260, 264, 266, 269, 280, Lankelma, H. P., 244 (42), 245, 866 (42, 285, 286, 288 (5),289 (18, 20) , 441 Johannson T., 261, 289 (9) Lstta, J. E., 130 (57), 156 (57) Johnson, E. A., 130 (601,165 (60) Leighton, P. A., 203, 205, 866 (6) Johnson, M .F. L., 88 (46), 90 (46) Lernieux, R. U., 81, 89 (32) Johnson, R. P., 166 (30), 196 (30) Lennard-Jones, J. E., 168, 171, 195, 198 Jones, H., 189 (34), 199 (135) Jones, I. H., 123 (53), 156 (53) Lewis,J. R., 9, 86 (9) Jones, J. P., 118 (47), 119 (47), 123 (47), Leypunsky, 0. I., 163, 196 (15) 164 (47) Liang, 5. C., 4, 9, 21, 23, 86 (88, b) Joyner, L. G., 73-76, 81, 89 (10) Lien, A. P., 226 (31), 228 (31), 231 (31), Jura, G.,75-79, 89 (13-15) 238 (311, 242, 247 (31), 866 (31) Jurgens, H.,171 (46), 197 (46) Linn, C. B., 37, 60,61, 62, 63 (19, 20),64 (42) K Lipson, H., 261, 289 (8) Livingston, H. K., 76, 89 (16) Kashtanov, P., 180 (76), 197 (76) Kassel, L. S., 214, 239 (13), 240 (13),266 Lonsdale, K., 261, 888 (7) Lovell, W. G., 224, 866 (29) (13) Luder, W. F., 31, 89 (7) Ka~anskii,B. A., 104,113 (22)
Horiuchi, H., 92, 119 (4) Horiuti, J., 171 (38), 181, 183 (85), 184, 193 (1251, 196 (38),197 (78, 851, 198 (87, 88, 125) Horn, O., 117 (26), 164 (26) Horne, W.A., 118 (45, 4?), 119 (45, 471, 123 (45, 47, 52), 136 (67,68), 164 (45, 47), 166 (52, 67, 68), 214, 865 (15) Horrex, C., 186, 198 (98) Hosemann, R.,264, 389 (14) Howard, J., 3 (see Taylor), 26 (6d) Howes, D. A., 118 (45), 119 (45), 123 (45), 164 (45) Huberman, N. D., 193, 198 (126a) Hughes, D. R., 191 (1121, 198 (112) Hulla, G.,28, 63 (2), 218, 244 (21), 256 (21) Hummer, R. W., 105, 113 (23) Hunsmann, W.,19, 20, 21, 26 (15)
295
AUTHOR INDEX
M McAlliter, S. H.,33, 38, 42, 56, 57, 58, 83 (12),224, 244, 266 (28) McAteer, J. H., 202, 214,966 (3) McReynolds, L.A., 47 (see Alden), 63 (30) Miirti, W.,111, 114 (37) Marschner, R. F.,41, 42,58, 59,63 (24) Marsh, J. D.F., 110,114 (33) Martin, F.,117 (32),143, 164 (32) Martin, H.Z.,130 (56),I66 (56) Mavity, J. M.,229,237,238,240,266(32) Mayrhofer, K.,165 (23b),196 (23b) Merkel, H.,138, 149 (74),166 (74) Messner, G.,21, 26 (16s) Meyer, K.,117 (20), 163 (20) Michaelk, L.,192,198 (123) Miles, C . B., 33,63 (13) Miller, A. R.,169, 196 (36) Milligan, W. O., 257 Moldavskii, B. L.,108,I14 (29),214,239, 966 (14),266 (33) Montgomery, C. W., 73-75, 81, 89 (lo), 202, 214, 266 (3) Morcom, A. R., 103, 113 (20) Morikawa, K.,13, 26 (12a, b), 181, 184 SO,^) 197 (so), 198 (90) Morrison, I. A., 81, 89 (32) Morrison, J. L.,161 (ll),196 (11) Murphree, E.V.,130 (56,62),166 (56,62) Myddleton, W.W.,141 (78),142 (SO), 166 (78,80) N Naah, A. W., 116,163 (5,6) National Research Council, 92 (3),223 (3) Nelson, J. B., 261,289 (8) Neniteescu, C.D., 209, 244,254,966 (lo), 966 (41) Neumsnn, R., 118 (46), 119 (46), 123 (461,164 (46) Newling, W. B. S.,110,114 (33) Nikiforov, V., 108, I 1 4 (27) Nieovkina, T., 214,239,266 (14),266 (33) Norris, J. F.,50, 64 (33) Nysewander, C. W.,244, 966 (39) 0
Oberfell, G. Q., 29, 83 (4) Oblad, A. G.,202,208,209,266 (2,9) Ogden, G., 3 (see Taylor), 26 (6e), 184 (87), 198 (87)
Okamota, G., 171 (38),I96 (38) O’Kelly, A. A., 30,63 (5) Otros, Y.W.,223 (see Beek), 966 (25a) Owen, J. R.,82, 89 (38)
P Padoa, M., 108, 114 (25) Palmer, W.G.,188, I98 (102) Pauling, L.,167, 195,196 (31) Pearsall, H.,31, 63 (6),217,266 (la) Peebles, W.C.,151 (91),166 (91) Perry, S.F.,224, 244, 966 (38) Phillips, N. E.,224, 266 (29) Phinney, J. A., 130 (631,I66 (63) Pichler, H.,117 (22,37),136,138,140 (77), 142, 143, 144, 145, 146, 147, 148, 149, I63 (22), 164 (37),166 (74, 77), I66 63-85) Pickett, G., 79, 89 (24) Pim, F. B., 38 (see Birch), 56 (see Birch) 59 (see Birch), 63 (21) Pines, H., 27, 43 (26),44,63 (26,27),204, 206-209, 214, 217-219, 221, 229 (32), 237 (32),238 (32),239,240,243 (7,23, 35),966 (7,8, 11, 13),966 (19,23,32, 35) Pitkethly, R. C., 189, 198 (108) Pitzer, E. C., 110, 124 (35) Pitzer, K.S.,32,63(g),213,239,240(12), 247,866 (12) Plate, A. F., 104 (22).113 (22) Polanyi, M., 92, I 1 3 (4), 171, 181, 183 (85,86), 184,186,193 (125,1%), 195, 198 (39), 197 (45, 77, 78, 85), 198 (86-88, 97, 98, 125, 129) Pollard, W. G., 171, 197 (44) Popkm, A. H.,242 (37),966 (37) Powel, A. R.,118 (49),119 (49),123 (49, MI, 164 (49,50) Powell, T. M., 216,242, 966 (17) Prosen, E.J. R., 32, 63 (9), 213, 239, 240 (12) 247, 956 (12)
Q
Quastel, J. H., 194, 199 (132) R Ramsbottom, J. E.,191 (115),198 (115) Randlett, H.E.,224, 244,866 (28) Raymond, C.L.,223,266 (26) Reder, R., 117 (22), 163 (22)
296
AUTHOR INDEX
Reichl, E. H., 118 (41), 119 (41), 123 (41), 133 (41), 135 (41), 164 (41) Reid, E., 216, 242, 266 (17) Reynolds, D. A,, 116, 269 (11) Reynolds, P. W., 191 (114), 198 (114) Ridcal, E. K., 24 (20), 26 (20, 93, 94, 101, 102 (17), 105, 113 (7, 8, 17), 114 (24), 150, 158 (891, 158, 159, 163, 164 (171, 166 (28), 167, 170, 176, 177, 180, 181 184-188, 190, 193, 194, 195 (1, 2), 196 (6, 17, 28, 33a), 197 (79), 198 (89, 94, log), 199 (130) Rics, H. E., 88, 90 (46) Riley, D. P., 281, 289 (8) Rittenberg, D.,193, 198 (126b) Ritter, €1. L., 86, 90 (43) Roberts, G.,Jr., 130 (63), 165 (63) Roberts, J. K., 158, 159, 161 (4, l l ) , 162 (4, 12), 165-167, 168, 169, 170, 172, 176, 183, 193, 198 (4, 5, 11, 12, 32, 361, 197 (50,70) Roelen,0.,117(23,34), 119(34),154(23,34) Roess, L. C., 80, 89 (30), 264, 267, 280, 989 (17, 19) Roginsky, S., 174, 197 (54) Roginsky, S. Z., 193 (128), 198 (128) Rosenfcld, L.,118 (45), 119 (45), 123 (45), 164 (45) Ross, W. E., 33 (scc McAllister), 38 (see hfcMister), 42 (see McAllister), 56 (see McAUister), 57 (see McAllister), 58 (see McAllister), 69 (12), 224, 244 (see McAlIister) 266 (28) 62) Rossini, F. D., 32, 40 (see G ~ ~ O W (see Glasgow), 63 (91, 84 (43, 213, 239, 240, 247, 266 (12) Rothrock, H. S., 242 (36), 256 (36) R u b Chemie, A. G.,117 (33),122,164 (33) Rummel, K. W.,174 (57), 197 (57) Rupp, E., 104 (181, 188 (18) Russel, W.W., 68, 89 (5) Ryan, P., 130 (581, 266 (58) S Sachanen, A. N., 30, 63 (5) S a c k , H., 174 (m), 297 (58) Sadek, H., 13 Schadc, H. A., 119 (341,164 (34) Schechter, A., 174, 197 (54, 55) Schindler, H., 118 (45), 119 (45), 123 (451, 164 (45)
Schlenk, F., 192 (119), 198 (119) Schmerling, L., 33, 34, 35, 37, 38, 40, 41, 42, 44, 45, 46, 48, 50, 51, 65, 69 (14, 16, 28, 29), 64 (34, 38), 218, 223, 226, 254, 256 (19, 30) Schneider, A., 34, 35, 39, 40, 42, 69 (15) 218,158 (22) Schroeder, W. C., 118 (42, 46), 119 (42, 46), 121 (42), 123 (42,46), 164 (42, 46) Schubmt, M. P., 192, 198 (123) Schuit, G. C. A., 203, 226, 239, 265 (4) Schwab, G. M., 189, 198 (106) Schwamberger, E., 165 (23b), 196 (23b) Sekora, A., 264, 989 (15) Selwood, P. W., 176, 197 (68) Sensel, E. E., 244 (421,252 (42), 254 (42), 258 (42) Sergienko, S. R., 104 (22), I13 (22) Sheriden, J., 95, 96, 119 (11) Sherman, A., 171 (37), 174 (62), 296 (37), 197 .(62) ShockIey, W., 171, ,197 (43) Shterner, R. A., 239, 956 (33) Shuikin, N. I., 108, 114 (27) Shull, C. G., 80,86, 87, 89 (30), 90 (44), 264, 267, 280, 289 (19) Sickmen, D. V., 3 (see Taylor), 4, 26 (6b) Smith, A. E., 2, 13 (4a), 28 (4a), 94, ff9 (9),150 (9), 165 (9), 166 (91,167 (9), 194 (9),196 (S), 259 (I), 288 (1) Smith, C. S., 261, 289 (12) Smith, D. F., 100, 113 (14), 116, 150, 159 (7, 11, 16) , Smith, E. A,, 3 ( R e Taylor), 24, 26 (6g) Smith, H. A., 80, 89 (31) Smith, S. L., 117 (35), 119 (35), 164 (35) Smith, W. R., 244 (42), 252 (42),254 (42), lei76 (42) Solomon, E., 264, 266, 285, 286, $89 (18) Starr, D.F., 108, 114 (26) Stearn, A. S., 194 (131), 196, 199 (131) Steiner, H. S., 189, 198 (108) Stephenson, M., 192 (124), 198 (124) Stevenson, D. P., 223 (see Beek), 266 (25a) Stewart, T. D., 38, 69 (22) Stolyarova, P. A., 108, 114 (27) ' Storch, H. H., 85 (41), 90 (41), 117, 12: (50), 130 (55), 141, 142, 169 (18), 164 (501, 155 (56)
297
AUTHOR INDEX
Streiff, A. J., 67 (see Glaegow), 62 (see Glasgow), 64 (40) Strobele. R., 192 (122), 198 (122) Strother, C. O., 3 (see Taylor), 4,9,26 (6c) Sturgia, B. M., 50, 64 (33) Surhmann, R., 163,196 (14) Sutherland, R. E., 223 (see Chenicek), 224 (27), g66 (27) Swearingen, J. E., 244, 266 (39) SeenbGyorgyi, A., 195, 199 (134) \
T Tait, T., 38 (see Birch), 66 (see Birch), 59 (see Birch), 63 (21) Tamm, I., 171, 197 (42) Taylor, H. S., 1, 3, 4, 7, 9, 13 (see Morikawa, we Kernball), 14, 21, 22, 23, g6 (2, 6a-g, 8a, 9, 12a, b, 13, 14), 103, 105, 113 (19, 23), 165 (22), 166, 168 (33b) 171, 174 (60, 62), 175, 176 (64), 181, 184, 189, 193 (127), 196 (22, 25, 33b,c), 197 (60,62, 64, &198 I), (90, 107, 127) Teller, E., 2,4,26 (5), 68-70,72,77,79,89 (7) Thau, A., 117 (30), 164 (30) Thiele, E. W., 82, 89 (36) Thomas, C. A., 244 (40),266 (40) Thompson, 0. F., 123 (53). 166 (53) Thompson, R. B., 47, 48, 49, 63 (31) Thurmond, C. D., 202,266 (1) Tokareva, K., 220, 266 (25) Treer, R., 264, g89 (15) Trenner, H. R, 13 (see Morikawa), 26 (12 b), 184 (see Taylor), 198 (90) Tropsch,.H., 116, 117, 163 (2, 3, 9, 10) Trotman, C. G., 191 (115),'198 (115) Tsuneoke, s., 116 (15), 141 (79), 163 (15), 166 (79) Turkevich, J., 103, 113 (19), 176, 189, 197 (68),198 (107) Twigg, G. H., 93, 94, 112, 113 (6a, b, 7), 114 (39), 167, 184-188, 198 (33a), 198 (89, 91-94, 96) man, C. W., 130 (56) 166, (56) Teukervanik, I., 220, 268 (25)
V VanCleave, A. B., 161 (lo), 196 (10) VanNordStrand, R. A., 88 (46), 90 (46) Verheua, J., 203, 226, 239, 966 (4) ViSeer, G. H., 95, 103, 104, 113 (10, 18)
W Wackher, R. C., 204, 206-209, 217, 219, 221, 229 (32), 237 (32), 238 (32), 240, 243 (7, 23), 266 (7, 8, ll), 266 (23, 32) Wagner, C. D., 223 (see Beek), 266 (25s) Walker, J. J., 141 (78), 142 (80),166 (78, 80) Walker, S. W., 130 (57), 166 (57) Wang, J. S., 169, 196 (35) von Wangenheim, 138 (70), 166 (70) Warburg, O., 191 (118), 192,198 (118,120) Ward, R., 273, 289 (22, 23) Warhurst, E., 186, 195 (133), 198 (99a), 199 (133) Warren, B. E., 262, 269, 271, 272,280, ,989 (21.24) Waugh, R. C., 226 (31), 228 (31), 231 (31), 238 (31), 242, 247 (31), 252, 266 (31,
44a)
Webb, G. M., 61,62,64 (43) von Weber, U., 117, 118, 164 (40) Weinberger, E. B., 73-75, 81, 89 (10) Weir, H. M., 135 (66), 166 (66) Weller, S., 142, 150, 152 (go), 166 (81, 90) West, H. L., 118 (451, 119 (45), 123 (45), 164 (45) Westgren, A., 108, 114 (31) Wheeler, A., 2, 13 (4a), 26 (44, 82, 86,87, 89 (37), 90 (42), 94, 113 (9), 159 (9), 165-167 (9), 194 (9) 196 (9), 259 (l), 288 (1) White, T. A., 13, 26 (ll), 67, 89 (6) Whitmore, F. C., 35,63 (17), 218, 242,266 (20, 36, 37) Whittle, E. L., 242 (37), 266 (37) Wiedeking, K., 117 (28), 164 (28) Wieland, H., 191, 198 (117) Wigner, E., 174, 175, 197 (59) Williamson, A. T., 3 (see Taylor), 7,26 (6a) Willingham, C. B., 57 (see Glasgow), 62 (see Glasgow), 64 (10) Wolfenden, J. H., 165 (20), 186 (20) Wooten, L. A., 81,89 (35)
Y Yurev, Y. K., 108, 110, 114 (28, 34)
z Zelinskii, N. D., 104 (22), 108, 113 (22), 114 (28), 189 (105), 198 (105) Zuffanti, S., 31, 63 (7)
Subject Index ' A Accommodation coefficient, 159-163 Acetyl chloride catalyst, 30 Acetylene, 95-97, 100, 188 Acetylenic bonds, 186 Acetylenic complex of ethylcne, 187 Acid catalysts, 28, 32, 38 Activation, biological, 191-193 Activation energy, of adsorption, 9 of adsorption and desorption, 20 of alcohols, 188 carbon-hydrogen and carbon-carbon bonds, 181 of carbon monoxide oxidation, 111 of chemisorption, 4, 8, 17 collision numbers, 16 decrease by catalysts, 194 decrease by enzymes, 194 of desorption, 24 of deuterium exchange, 24, 183 differcnt magnitudes, 7-26 electron concentration, 189 ethylene hydrogenation, 186 formic acid dehydrogenation, 189 of hydrogcn adsorption, 7, 14, 20, 107 of hydrogen chemisorption, 22 of hydrogen desorption, 7, 20 of hydrogen on metals, 166 of hydrogen on tungsten, 178-179 of parahydrogen conversion, 183 recombination of hydrogen atoms, 174 transition from van dcr Waals adsorption to chemisorption, 15 Activation entropy, 194 Active centers, 2, 3, 20 Active spots, 193 Adsorbed films,76, 170-173 Adsorbed gases, 171 Adsorbed molecules, geometric arrangement of, 77-78 Adsorbed particles, 168-170 Adsorbents, activation energies, 7-26 capillaries of, 73-75 monolayer on, 67-89 surface areas, 7-26
Adsorption, see also Heat of adsorption, Chemisorption, van der Waah adsorption of argon, 87-88 of butane, 87-88 capacity of nickel catalysts, 19 of carbon dioxide, 66 of carbon monoxide, 110 of cesium, 171 on chromium oxide gel, 6 complexes, 95 on copper-magnesium oxide, 9 of deuterium, 2 direct measurement, 165 dissociative, 170-1 7 1, 178-191 energy, 171, 178-191 of ethylene, 94 gaps, 180, 194 of hydrogen, on chromium catalysts,6-7 on copper catalysb, 9 on cracking catalysts, 107 on iron ammonia catalysts, 11-13 on manganese-chromium oxide, 6 on manganous-chromium oxide, 7 on metals, 166 on nickel catalysts, 13-21, 165 potential energy curve, 171 on tungsten, 21, 85, 162, 165-168 van der Waals,4, 13 on zinc oxides, 4, 6-7 hysteresis, 87 isobars, 4-25 isotherms, 66-89 measurements, 5-26, 77-78 on metals, 166 Multiplet theory, 91 of nitrogen, on carbon, 73 on charcoal, 80 low temperature, 67-89 on nickel-kieselguhr catalyst, 16 on porous glass, 87-88 on tungsten, 21-22, 85, 166 on zinc oxide, 6 of olefins, 39 of oxygen, 162, 166, 176-177 298
299
SUBJECT INDEX
of palmitic acid, 80-81 of parahydrogen, 2 promoter distribution, 82 rate of, 4-26, 166 reversible, 39 on silica gel, 69-71 of stearic acid on solids, 80 surface, 171 surface complex, 21 and surface heterogeneity, 9 on titanium oxide, 6 of trimethylethylene, 94 on tungsten, 21-22, 85, 162, 165-168, 171
two-layer reaction, 158 two-point, 91-112,190 van der Waals, 2-4, 13-22,66-89 of water on titanium oxide, 6 Adsorption-desorption of hydrogen, 20-21 Alcohols, 59, 101, 115, 188-191, 136 Aldehyde formation, 100, 135 Alkanes, isomerization of, 201-256, 8ee Isomerization of the various alkanes Alkazid process, 119 Alkylates, 28, 38,40,47-49, 54,56-57, 60-62
Alkylating agents, 29, 38, 56, 59, 62 Alkylation, of n-butane, 53 catalytic, 30-32 destructive, 31, 39 -43, 45, 57, 59, 61-62 of ethylene, 27 of hexanes, 27, 43, 57 of iaobutane, 30-62 with t-amyl alcohol, 59 t amyl chloride in, 40 with 1-butene, 31, 36-38, 42, 44,50,56 with Zbutene, 31,35,37-38,42,44,50, 56-57
with 8-butyl fluoride, 37, 62 with l-butyl chloride, 40 with butylenes, 54-58, 60 carbon-carbon cleavage, 40 carbonium ion, 36, 40,42 catalysis, aluminum bromide, 45, 53 aluminum chloride, 37-40, 43-46, 50-54
aluminum chloride-monomethanohte, 37, 42, 50, 51
aluminum chloride-nitroalkane, 51-52
aluminum chloride sludges, 47 boron fluoride, 54 hydrogen fluoride, 37, 60-62 sodium chloroaluminate, 52 sulfuric acid, 38, 41, 42, 55-59 tritium sulfuric acid, 38-39 zirconium chloride, 55 chain mechanism, 40 chloroalkylation, 45 destructive alkylation, 40, 61 with diisoamylene, 58 with diisopropyl sulfate, 59 with ethyl bromide, 53 with ethylene, 30-33, 39-40, 4344, 47, 52, 54-55
with 2ethyl-l-hexanoll 58 hydrogen exchange, 40 hydrogen transfer, 51, 54 with isoamylene, 41 with isopropyl chloride, 44-45, 52 with isopropyl fluoride, 60, 62 methyl bromide in, 53 with Zmethyl-2-butene, 58 methylene transfer, 40 with Zpentene, 42, 58 presaure effect, 54 products, 30-33, 3545, 47, 50-62 with propene, 30-32, 39-40, 44, 46, 50-52,55-57,60-62
with propylene trimera, 53 promoters, 44,48, 51 proton transfer reaction, 42 temperature effect, 44-45,
50-51,
54-56,6041
trimethylethylethylene in, 40 of isoparaffins, 27-63 with alcohola, 59 with alkyl chlorides, 44-45, 59 with alkyl hydrogen sulfate, 58-59 alkyhte, 28,38,40, 4749, 54, 56-57, 00-62
alkylating agent, 29, 38, 56, 59, 62 aluminum-chlorine ratio in, 49 autodestructive alkylation, 228, 255 of benzene, 50 carbonium ion, 36-37 catalysts, aluminum bromide, 28,45,63-54
3 0
BUBJECT INDEX
Alkylation (continued) aluminum chloride, 27, 28, 34, 37-40. 43-46, 49, 54
aluminum chloride sludges, 31, 4650
aluminum chloride-monomethano late, 31, 37,42,&51 aluminum chloride-nitroalkane, 3132, 51-52
boron fluoride, 27-28, 39, 54 complex, 43 homogenous, 30 hydrogen fluoride, 28,32,34,37-38,
w, 43,5982
hydrogen fluoride sludges, 60 sodium chloroaluminate, 52 sulfuric acid, 28, 32-34, 38, 40-43, 55-60
sulfuric acid sludges, 55 zirconium chloride, 28, 54-55 chain mechanism, 37, 4 3 4 4 chloroolefins, 4 5 4 6 dealkyltrtion, 40 depolymerization, 40 destructive alkylation, 31, 39-43, 45, 57, 59, 61-62
disproportionate rcactions, 54 ester formation in, 43 hydrogen disproportionation, 42 hydrogen transfer, 4144, 51, 54, 57-59, 62
intermolecular hydrogenation, 44 mechanism, 32-39 with olefins, 43-44, 55-63 polymerization of olefins, 42 promotem, 27-28, 30-31, 38, 4144, 48-54
destructive alkylation, 59, 62 hydrogen transfer, 57, 62 with kopropyl chloride, 45, 52 with isopropyl fluoride, 62 products, 42, 45, 52, 57, 59, 62 with propene, 61 of naphthenes, 27 of propane, 29-30 selective, 01 thermal, 29-30 Alloxazine, 192 Alloys, Brillouin zone, 189 Ally1 chloride in chloroalkylation, 45-46 Alumina, alphrr, 281-285 catalysts, 101, 137, 259, 281-285 exchange of heterostom on, 110 heterocyclic compounds, 110 hydroforming Of, 281-285 aurface area, 281-285 x-ray pattern, 259 conversion of, 103 gel, 285-288 influence of molybdenum trioxide on, 280-288
low soda, 281-285 particle sizes, 280-288 promoter, 82-83, 111. support for catalysts, 103, 216, 223 surface area, 281-285 temperature effect on, 280-288 y-Alumina, 104, 281-285 Aluminum bromide catalyst, in alkane isomerization, 202-203, 207211, 215, 219-221, 244, 249, 252-253
223, 225-227,
in butane isomerization, 202-203, 207self-condeneation, 44 211, 215, 219-221 side reactions, 39-43 in heptane isomerization, 252-253 sludge, 43, 46-50, 60 in h e m e isomerization, 244,249 Whitmore’s theory, 35 in isoparaflin alkylation, 28, 45, 53-54 of isopenhe, in pentane isomerization, 225-227 with alcohol, 59 in propane isomerization, 223 with 2-buteneI 57 reaction with deuterium oxide, 222 with I-butyl chloride, 59 reaction with oxygen, 220-221, 219 with butylenes, 42, 68 reaction with water, 209-211, 221-223 catalysts, aluminumchloride-nitrometh~ne,52 Aluminum bromidedeuterium bromide catdpt, 218-219 hydrogen fluoride, 40, 59 Aluminum bromide hexahydrate, 211 sulfuric acid, 40, 57, 59
SUBJECT INDEX
301
chemisorptioa of carbon monoxid eon, 83 chemisorptioa of hydrogen on, 4 deaorption-readsorption on, 11-12 exchange reaction, 183 Fixed Nitrogen Laboratory catalyst, 13 metallic iron in surface of, 84 nitrogen activation on, 25 in parahydrogen conversion, 182-183 promoted, 12, 82-83 rate of adsorption on, 12 surface measurement, 67 Ammonia-deuterium exchange reaction, 182-183 Ammonia synthesis, 25 Ammonium cupromolybdate, 101 &Amy1alcohol, 69 &Amy1chloride, 40 50-54 Angle of scatter, in isobutane chloroalkylation, 46 Geiger counters in measurement of, in isoparaffin alkylation, 27-28, 31, 34, 200-268 37-40, 43-46, 49, 54 Geiger-Mueller measurement of, 265-267 in pcntane isomcrizution, 224-225, 228, Guinier method for measuring, 269-271 231, 236-238, 241, 244 particle sizes, 263-264 promoted with water, 212-213, 244 photographic measurement of, 265-267 supported on alumina, 216, 223 principle, 257-258 supported on charcoal, 216 silica gel, 288 Aluminum chloride-charcoal catalyst, 207, small angle, 262-263 209, 216 techniques of measuring, 264-272 Aluminum chloride-hydrogen chloride theory, 259,264 catalyst, two-crystal method for measurement, antimony chloride in, 224, 244 269-271 in butane isomerization, 203-206, 217, x-ray spectrometer measurement of, 268 223 in x-rays, 258-273, 285-288 in isoparaffin alkylstion, 28, 31 Antimony trichloride, 224, 244 in pentane isomerization, 224-225, 22% Argon, 68, 87-88 238, 241-243 Aromatization catalysts, 101, 104, 105 Aluminum chloride-monomethanolate Atom, catalyst, 31, 37, 42, 50-51 carbon, secondary, 39 Aluminum chloride-nitroalkane catalysts, tertiary, 28, 32, 37, 57-58, 118 31-32, 51-52 quarternary, 118 Aluminum chloride-nitroparaffin comdeuterium gwhydrides, 181-184 plexes, 51 displacement of metal, 94 Aluminum chloride sludges, 31, 46-50 distances, 93 Aluminum oxyhalide, 219 exchange on nickel catalyst, 181 7-Aluminum trioxide, 103 hydrogen, 93, 98, 172-174, 181 Ammonia catalyst, metal, 94 activation energy, of deuterium exrecombination, 173-174 change, 183 spacing, 94 adsorbent of hydrogen, 11-13 Atomic-molecular hydrogen interaction, adsorption-desorption of hydrogen on, 21 180 Aluminum bromide-hydrogen bromide catalyst, 28, 204-205, 219, 246, 253 Aluminum chloride-alkali metal chloride double saltq 52-53 Aluminum chloride-bauxite catalyst, 224 Aluminum chloride catalyst, in alkane isomerization, 202-203, 207 209, 215, 219, 223-225, 228, 231, 236-238, 241, 244-245, 252 hntimony triclloride in, 224, 244 in butane isomerization, 201-203, 207, 209, 212, 215, 219, 224 in ethylene alkylation, 27 in heptane isomerization, 252 in hexane alkylation, 27 in h e w e isomerization, 245 in isohutane alkylation, 37-40, 43-46,
302
SUBJECT INDEX
Atmospheric pressure procesa, 118 Atmospheric pressure reactors, 120-121
B Barium oxide-ferric oxide catalyst system, 273-279
Barium carbonate-ferric oxide cat,alyst, 273-277
Barium sulfate, 77 Bent crystal monochromators, 262-263 Benzal chloride catalyst, 30 Benzene, C-C distance in, 92 cracking inhibitor, 229, 231-232, 234, 236, 240-243, 249
from cyclohexane, 9 1 1,4dimethylbenzene, 237 in pentane isomerization, 232, 234 Benzene-deuterium exchange reaction, 186 Benzol, 116 Benzyl chloride catalyst, 30 Biological oxidations, 191-192 Boltzmann factor, 8 Bond angles, 94, 112 Bonds, C-C, in Fischer-Tropsch synthesis, 98 C C in lattice of catalyst, 140 C-C rupture of, 107 C-Co in Fischer-Tropsch synthesis, 98 C-H rupture of, 188 C-metal, 93-95 C-Mo, in destructive hydrogenation, 106 double, migration of, 93, 98, 185 ethylenic and acetylenic, 186 Bond type of adsorbed films, 170-173 Boron fluoride, 203-204 Boron fluoride catalyst, 27-28, 39, 54-55 Boron fluoride-hydrogen fluoride catalyst, 31
Brabag catalyst, 127-128 Brillouin zone of alloys, 189 Bromine, 209 Bromoparaffins, 203-204 Brunauer-Emmett Technique, equqtion, 68-71 advantzges of, 72-73 comparison with other methods, 72 derived by statistical thermodynamics, 79 linear plot, 71
modification f& capillaries, 73-75 modification for condensation, 75 method, 3, 4, 25 compared with Harkindura, 78 plots, 71 present status, 88-89 surface area measurement, 3, 4, 25 theory, 70 Bulk carbide, 150-152 Butane, adsorption isotherms, 68 adsorption on porous glaas, 87-88 adsorption on silica gel, 72 adsorption in surface measurement, 2,72 isomerization of, 201-224 (see ulso Isomerization of butanes) molecular c r m section area of, 78 Butane-deutcrium cxchangc, 219 Butanol, 220 Butene, catalyst surfacea in dehydrogenation of, 82
double bond migration in, 185 product of thiophene decomposition, 110 1-Butene in alkylation, 31, 36-38, 42, 44, 50, 56
ZButcne in alkylation, 31, 35, 37-38, 42, 44, 55, 56-57
Butene-deuterium exchange, 93 ButoxyaIuminum dihalide, 220 Butyl alcohol, 188 n-Butyl alcohol, 136 8-Butyl alcohol, 42, 59 tButyl alcohol, 42, 59 t-Butylbenzene, 237 s-Butyl carbonium ion, 218 I-Butyl ctrrbonium ion, 54 t-Butyl chloride, 34, 38, 40, 42, 46, 59 8-Butyl fluoride, 37, 62 Butylene-deuterium exchsngc, 185 n-Butylene, 42, 55 C
Cameras for x-ray diffraction, 260 Capillaries, B. E. T. equation, 73-75 Capillary condensation, in pores, 86-88 Capillary condensation, theory, 74 Carbanion, 33, 37 Carbide, catalysts, 25 formation on iron catalysts, 138-140
SUBJECT INDEX
Carbide (wnlinued) intermediate hypothesis, 152 theory, 150 Carbidemethylene polymerization hypothesis, 153 Carbided cobalt catalysts, 160-152 Carbidic carbon, 138-140 Carbon, adsorbent of nitrogen, 73 a t o m , 28, 32,37, 39,57-58, 118 deposition on iron catalysts, 138-140 deposition in synol process, 135 radioactive, 139, 152 Carbon black, surface area, 79 Carbon-carbon bonds, 98, 107, 140 Carbon-carbon cleavage, 33,40 Carbon-cobalt bonds, 98 Carbon dioxide, 66,68, 84 Carbon disulfide, hydrogenation of, 109110
Carbon isotopes, 85 Carbon-metal bond, ethylene on nickel, 93-95 Carbon-molybdenum bond, 106 Carbon monoxide, activated adsorption of, 110 adsorption isotherms, 68 aldehyde formation, 100 attachment to oxidation catalyst, 11 1 catalytic oxidation, 110-111 chemisorbed, structure of, 167 chemisorption of, 25, 66,83 cobalt catalyst in hydrogenation of, 97 femc oxide catalyst in oxidation of, 111 hydrogenation, 97-100 in isosynthesis, 101 oxidation, 273-279 poisoning in parahydrogen conversion. 179 Carbon-nickel distance, 95-96 Carbon-palladium distance, 96 Carbonium ion, in akylation, 33,35-37, 40, 42 8-butyl-, 218 ethyl-, 40 in hrnenzation, 240-243 n-pentyl-, 242 tpentyl-, 242 Carbonyl formation, 143, 149
303
Catalyst, general active points of, 82 activity of, 82 complex formation of, 43 crystallites in, 259, 280-288 Mraction patterns, 258-260 geometric factors in surfaces of, 91-1 13 heterogeneity of surfaces, 25-26 industrial, 193 isotopic molecules on, 24 lattice dimensions, 92 life, 144, 148-149, 231, 236 particles in, 280-288 poisoning of, 2,91, 111, 113; 194 porosity of, 82 potential barriers lowered by, 195 pressure effect on, 140, 143 promoters, 2, 12, 17,48,54,82-83, 110111, 116 structure from x-ray scattering, 257-288 surface, geometry of, 193-194 surface meaaurements, 65-89 Catalysts, specific, see under specific namea Catalytic activity, of chromium oxide gel, 103 of components of BaO-FenOS system, 277-279 crystallographic orientation of films, 94-95 of heteroatoms, 110 of hydrogen, 157-195 and spacing of adsorbed molecules, 94 specific, for crystal facee, 2 surface area, 82 vdency electrons, 195 and x-ray analysis, 257-288 Catalytic alkylation, 27-60 Catalytic exchange, heterocyclic compounds, 110 Catalytic reactions, 25, 92,95 Cementite, 139, 280 Cesium adsorbed on tungsten, 171 Chain mechanism in catalytic alkybtion, 33, 37, 40-41, 4344 in isomerhation, 217, 223, 240-242 in thermal alkylation, 29-30 Chain reaction, acetylene to higher hydrocarbons, 97
304
BUBJECT INDEX
Charcoal, alumhum chloride supported on, 218 &B carrier, 207 catalyst for parahydrogen conversion,
equilibrium values, 23 yform, 103 heats of adsorption for hydrogen on, 23 n-heptane cyclization, 189 174-176 hydrogendeuterium exchange on, 175 nitrogen adsorption on, 80 in parahydrogen conversion, 176 outgsssing of, 176 in ring closure reactions, 101 surface detarmination of, 80 supported on alumina, 103 Charcoal-alumina chloride catalyst, 209 surface area, 104 Chemisorbed film, 188, 168-173, 193 Chromium oxide gel catalyst, Chemiaorption, adsorptiondesorption-readsorption of activation energy of, 4, 8, 15-17 hydrogen on, 6-7 area measurement, 66,85 catalytic activity, 103 of carbon dioxide, 84 chemisorption of hydrogen on, 22-23 of carbon monoxide, 25, 83 inactivation of, 103 on clean metals, 158 Chromium oxide-iron oxide catalyst, 117 differentiated from van der Waah Cis-addition, in hydrogenation, 188 adsorption, 2-3 Cobalt, cubic and hexagonal, 152 ofethylene, 167, 185, 187 Cobalt carbide, 97, 149, 188 gaa molecules and r-electrons, 167 Cobalt carbonyl formation, 148 heat of, 13, 19, 23, 187 Cobalt catalyst, of hydrogen, adsorption of ethylene, 94 on ammonia catalyst, 4 alkali-activated, 115 on chromium oxide, 22-23 carbided, 150-152 heat of, 13, 19, 23, 187 carbon monoxide hydrogenation on, g7 on iron, 11-13 in Fiecher-Tropsch synthesis, 94,97-100, on metals, 168-184 140, 144, 187 nature and packing, 167 life, 144, 148 on nickel, 13-19, 187 0x0 synthesis, 100 structure, 167 products from, 117 in5uence on van der Waals adsorption, promotem of, 116 81 spacing of atoms, 94 a~ influenced by Boltzmann factor, 8 surface covering by hydrocarbons in, 99 layer interaction, 193 surfsce area, 85 of nitrogen on tungsten, 22 CobaIt hydrocarbonyl catalyst, 138 of oxygen, 167 Cobalt-manganese oxide-copper catalyst, surface meseurementa, 66 116, 150 Cobalt oxide catalyst, 85 temperature regions, 3 on tungsten, 22, 159, 161 Cobalt oxides, higher, 110 Chloroalkylation of isobutane, 46-18 Cobalt-thoritt-magnesia-kieselguhr cataCblorobencene, 237 lyst, 117-120,123, 133, 135, 142-143, l-Chloro3,3dimethylpentane,34, 40 146, 150-161 L - C p l o r o 3 , ~ i r n e t h y ~ ~ n33, t ~ 46 el coenzymes I and 11, 191-192 Coke oven gas, 119 ZhIoroform catalyst, 30 Collision efficiency of paramagnetic sub2hloroolefh, 33,4646 2hromium oxide catalpt, stance~,175 Collision numbers, 16 activation energy of, 22 Condensation of adsorbed gas in capillaries! on -y-Al,Oh 108 aromathation catalyst, 103 74 chemimrption of hydrogen on, 22-28 Condenaation of kbutyl chloride, 34
UBJECT INDEX
Zondensation of isobutane, 46 Contact potential, check for adsorbed oxygen, 175 measurement, 163 nickel-nickel-oxide, 165 for tungsten, 164, 176-177 Contractions in hydrocarbon synthesis, ,
147
Converters, in Fischer-Tropsch plants, 120-121 in fluidized iron proceas, 1W131 multi-stage operation of, 121 Copper, 116 Copper catalyst, adaorption of ethylene on, 94 adsorption of hydrogen on, 9 copper nitrate aa a source, 9 desorption-readsorption effects, 11 in hydrogenation of olefins, 94 hydrogen overvoltage on, 174 nickel as a component, 94 recombination efficiency of hydrogen atoms on, 174 spacing of atom, 94 surface areas, 9 thorium nitrate as Component, 9 ?opper chromite catalyst, 136 2opper-iron catalyst, 116 2opper-magnesium oxide catalyst, 9 2opper oxide, 110 2opper phthalocyanine, 175 & m d u m , 281-286 ;racking catalyata, 106-107, 206,288 :racking inhibitors, aromatic hydrocarbons, 229-237 benzene, 229, 231-232, 234, 236, 240243,249 chlorobenzene, 237 cyclohexane, 237-238, 240,,247 cyclopentane, 247 1,4-dimethylbenzene,237 diphenyl ether, 237 in heptane isomerization, 254-255 in hexane isamerization, 247,249 hydrogen, 225-228, 240, 243, 247, 254-255 hydrogen chloride, 206-207, 225-226, 247, 265 isobutane, 240 naphthenes, 237-238, 249
305
organic additives, 225-226,229-238,247 in pentane isomerization, 225-238, 240-243 tbutylbenzene, 237 toluene, 237 Cracking suppressors, 8ee Cracking inhibitom Cracking reactions, 112 Crystallite size, 262-263 Crystallita in catalysts, 259,280-288 Crystallographic orientation of nickel films, 95 Cyclization, 189-190 Cyclohexane, 91,237-238, 240,244,249 Cyclohexene, 100 Cyclopentadiene polymerization, I10 Cyclopentane, 247-249, 251 Cytochrome reductase, 192
D Dealkylation, 40 Debye-Scherrer diagram, 261-262, 279, 282-284 Debye-Scherrer theory, 261-262 Decomposition effect on hexane equilibrium, 251 Dehydrogenase, 191, 194 Dehydrogenation, of Jcohols, 188-189 in bidogical oxidations, 191 of cyclohexane, 91 dcsorption in, 188 of formic acid, 189 metal d o y catalysts, 189 Depolyalkylation, 68 Depolymerization, 40 Desorption of hydrogen 7, 19-21, 24 of hydrogen deuteride, 24 in hydrogenation, 188 rate of, 4-26 Demrption-readsorption, on copper catalyste, 11 effects, 4-20 on iron-ammonia catalyste, 11-13 mechanism of, 8 on nickel catalysts, 13, 18, 20-21 on tungsten powder, 21 on einc-chromium oxide, 23 Deuteride surface on tungsten, 178
306
SUBJECT INDEX
Deuterium, activation energy of, 24,183 adsorption of, 2 exchange reaction with, ammonia, 182-183 benzene, 186 butane, 219 butene, 93 butylene, 185 ethane, 181 ethylene, 91-95, 181, 184-187 hydrides, 181-184 hydrogen, 23-24, 94, 175, 177-179, 185, 193, 219, 222 methane, 181 propene, 93 propylene, 185 unsaturated hydrocarbons, 184-187 water, 181, 183, 193 hydrogenation of methyl oleate with, 188
isotope, 2 Deuterium bromide-aluminum bromide ~&dyst, 218-219 Deuterium oxide, 181, 221-222 Deuteroxyaluminum dibromide, 221-222 Diatomite, 95 l,l-Dichloro-3,3-dimethylbutane,46 l,%Dichloroethylene, 46 1,2-Dichlor~4,4dimethylpentaneJ 33, 45 Diene synthesis, 195 Diffraction patterns of catalysts, 258-260 Diffusion, 166, 168-170 Diisopropyl sulfate, 59 Dimethyl ether, 101 1,4-Dimcthylbenzene, 237 2,2-Dimethylhutane, 31-32, 35,43, 49, 64, 138, 245-251, 254 2,3-Dimethylbutane, 30-34,40,43,46,4950,54,57, 138, 245-248, 26U-251,254 2,,5Dimethylhexane, 44 S,CDimethylhexane, 36 2,%Dimethylpentane, 31-32, 44, 50-51, 57, 61 2,4Dirnethylpentane, 44 3,3-Dimethylpentene, 116, 253 2,2Dimethylpropane, 238-239, 242 Diphenyl ether, 237 a, a'-Diphenyl-fl-picryl hydrazyl catalyst, 176
Disproportionate reactions, 54 Dissociation of hydrogen, 172-173, 181 Double bond, migration of, 93, 98, 186
E Elaidic acid, 187 Electric contact potential of metals, 159 Electron concentration in alloys, 189 diffraction, 95, 164 gas on metals, 171 unpaired, in adsorbed acetylene, 96, 97 valency, 196 r-Electrons, 167 Energetics of the transition state, 194-195 Enzymes, 192, 194-195 Equilibrium concentration, of butanes, 215 ofhcxanes,247-248 of pentanes, 238-239 Equilibrium constants, of butanes, 214-215 of hexanes,' 248-251 of pentanes, 238-239 Equilibrium isobars, 6, 14 Equilibrium state, 202-203 Equilibrium studies, 213-215, 238-240, 247-252 Equilibrium values of hydrogen adsorption, 23 Eachm'chiu coli, 192-193 Ethane, 95, 227, 181 Ethane-deuterium exchange, 181 Ethyl alcohol, 188-191 Ethyl bromide, 53 Ethyl carbonium ion, 40 Ethyl chloride, 48 %Ethyl-1-hexanol, 58 Ethylene, aa acetyledc complex, 187 activation energy, 186 adsorption, 94, ,185 aldehyde format.ion, 100 in alkylation of isobutanq30-33, 39-40, 43-44, 47, 52, 54-55 in alkylation of hexanes, 43 atom distances, 93 carbon-metal bond to catalyst, 93-95 chemisorbed, 167, 185, 187 condensation, 40
307
SUBJECT INDEX
configuration of 96 dietortion of bond angles 011 nickel 94 exchange reaction with deuterium 91-95, 181, 184-187 d&hylene, 185 tetradeutero 93 in Fischer-Trupech syntheaia, 100, 116, 160,188
geometrical poeition on nickel lattice, 93 heat of chemisorption on nickel, 187 from hydrogen and acetylene, 95 hydrogenation of, 92-96, 180, 186-187 on metal films, 187 se poison, 179 aa retarder, 94 esturation of, 96 spacing, 93-96, 185 Btructure of, 167 and synthesis gaa, 116, 150 in thermal alkylation, 30 Ethylenic bonds, 186 Evering, d'ouville, Lien and Waugh mechanism, 242-243 Exchange Reactions, in catalytic alkylation, 38-40 deuterium-ammonia, 182-183 deuterium-benzene, 186 deuterium-butane, 219 deuterium-butene, 93 deuterium-butylene, 185 deuterium-ethane, 181 deuterium-ethylene, 91-95,181,184-167 deuterium-h y drides, 181-184 deuterium-hydrogen, 23-24, 94, 175, 177-179, 185, 193, 219, 222 deuterium-methane, 181 deuterium-propene, 93
deuterium-propylene, 185 deuterium-unsaturated hydrocarbons, 184-187
deuterium-water, 181, 183, 193 ethylene- ddethylene, 185 ethylene-tetradeutero, 93 higher olefine, 93 hydrogen atoms, role in, 93 hydrogen-chlorine, 44 hydrogen-halogen, 34, 39-40 53 hydrogen-tritium 39, 216-217 in isomerieation of butnnea, 222-223 h t o p i c molecules on catalysta, 24
methaneethane, 181 migration of double bond in, 93
F 7-Ferric oxide, 104 Ferric oxide catalpt, 111 Ferric oxide-barium carbonate catalyst, 273-277
Ferroue carbide, 280 Filma, chemhorbed, bond type, 170-173 mobility of, 168-170 nature. of, 166 stability of, 193 empty s i t e in, 170 iron, 2 metallic, 187, 194 nickel, 2, 13, 159 oxide, 112 palladium, 2 platinum, 2 rhodium, 2 sulfides, 112 surface spreading force of, 76 tungsten, 159, 165, 178 Fischer-Tropsch aatalpts, area, 85 carbide, 25 cementite, 130, 280 cobalt, 9S99, 116, 187 cobaltmanganeae oxide-copper, 116,150 CobalbthOri&kidgh, 142-143, 145, 151
cobalt-thoriwnagneabkieaelguhr, 1 17, 150 cracking of hydrocarbon8 on,98 iron in, 97-100, 118-117, 187, 279-280 iron carbide in, 149,280 iron oarbonyl, 124 iron-wpper, 118, 150 magnetite in, 280 nickel, 97-100 nickel-manganese dioxidealuminnkieselguhr, 117 ruthenium catalyst, 146-148 8urfr~Wactivity, 85 x-ray stud&, 27!3-280 Fischer-Tropsch synthesis, acetylene, 100, 188
308
EUBJECT INDEX
Fischer-Tropsch synthesis (eonlin?cmi) aliphatic hydrocarbons in, 99 bonde, 98 bulk carbide, 151-152 carbide intermediate hypothesis, 152 carbidemethylene polymerization hypothesis, 153 carbide theory, 150 carbon isotopes, 85 earbonyl formation, 149 chemisorption of carbon monoxide, 25 cobalt carbide surface formation, 188 cobalt carbonyl formation, 148 ethylene in, 100, 116, 150, 188 F. Fischcr’s hypothesis, 149-1rfl fluidized-iron process, 130-133 forination of macromolecules, 98 gnsoline, 117 Hall hypothesis, 150 hydrocarbons, 25 intermediate pressure synthrsis, 145 intermediates, 97, 138-140, 14!) isomer formation, 98 kinetics, 140-149 mechanism, 149-153 methane formation, 98 methyl side groups, 98 methylene groups, 188 migration of double bond in, 98 nitrogen dilution effect, 140-142 olcfins in, 98 paraffins, 98 plant converters, 120 pressure coefficient, 140-149 pressure effect, 140-143, 148 products in, 116 radioactive carbon, 139, 152 ratio of gases, 118 reaction mechanism, 140-149 reaction rate, 140-142 reluted processes, 100 space-time yield, 132-133 temperature coefficient, 140-149 temperature effect on throughput, 141145 tempcruture effect on yield, 148 throughput effect on yield, 145-147 Fixed Nitrogen Laboratory catalyst, 13 Formic acid, 189, 199 Furan, 107-108 ,
G
Gaps in adsorption layers, 180, 184 GRSCOI, 121-122 Geiger counters, 266-268. Geiger-Mueller measurement, 263, 267 Gels, 80 Geometric factors in catalysts, 91-113 Glass,porous, adsorption on, 87-88 Glam spheres, surface area of, 80 Glucose 6 phosphate, 192 Glutaric acid, 194 Gold catalyst, 191 Gregg’s method, 77 Guinier method, 269-271
H Hall hypothesis, 150 Harkins end Jura method, 75-77 compared wit,h B. E.T. method, 76 Heat of adsorption, B.E. T . theory, 70 different magnitudes on one adsorbent, 7-26 of gases on mekls, 159 of gases on tungsten, 161-163 of hydrogen on chromium oxide, 23 of hydrogen on manganous-chromium oxide, 7 of hydrogen on nickel, 19-20 of hydrogen on tungsten, 162, 167 methods of measurement, 161-162 of oxygen on tungsten, 1132 timc effect in evolution of, 21 theory for interaction, 169 Heat of liquefaction, 71 IIcat losses on tungBten wire, 160 Heat of wetting on titanium dioxide, 76 Hcdman mechanism, 215-216 Hematin, 175 Hematoporphyrin, 175 Hemin, 175 Heptane-heptene ratio in cyclization, 189 n-Heptane, 78, 189, 190 Heptanes, isomerization of, 202, 252-255, see also Isomerization of Heptanes Heptene, 190 Heteroatoms, 110 Heterocyclic compounds, 107-1 10 Heterogeneity of oxides, 3-9, 22-26
SUBJECT INDEX
Heterogeneity of surfaces, and activation energy of adsorption, 9 and activation energy of chrniisorp! ion, 8 and catalyst preparation, 26 and catalyst variability, 25 change, by sintering, 16-17 influence on catalytic reactions, 25 iron, 11-13 metals, 9-22 nickel catalysts, 13-20 oxides, 3-9, 22-26 solids, 1-26 tungsten, 21, 168 zinc oxide, 24 Hexanes, equilibrium study of, 247-252 isomerization of, 202, 244-252. S P B nlno Isomcrization of hcxanes Hexahydrobenzaldchyde, 100 Hoesch-Benzin plant, 121, 125 Homogenous catalysts, 30 €Iopcalite catalysts, 110 Hot-gas recycle process, 123-124, 135 Hydride surface on tungsten, 178 Hydrides-deuterium gm exchnngr, 181-184 Hydrocarbons, aliphatic, Q9,101-106 aromatic, 101-106,232-237 attachment to molybdenum disulfide plane, 107 attachment to spinel lattice, 104 branched, by acetylene polymerization 97 catalysts for ring closure of, 101 chain growth in Fischer-Tropsch synthesis, 97-98 cracking of 98,106 destructive hydrogenation of, 106 exchange reaction with deutrrium, 184-187 molecules, zig-zag form, 99 space time yield, 147 synthesis, 25, 97-101, see also FisclirrTropsch synthesis unsaturated, 184-187 waxy, 100 Hydroforming with alumina catalyste, 281-285
309
Hydrogen, acceptors, 191 Rclivation energy, 7, 14, 20, 22, 107, 166, 174, 178-179 adsorption, capacity for nickel catalysb, 19 on chromium catalysts, 6-7 on copper catalysts, 9 on cracking catalysta, 107 and desorption on ammonia catalyst, 21 and desorption on nickel, 20 dircct measurement, 165 dissociative, 170-171, 178-191 energy, 171 equilibrium values of, 23 heat of, 19-20, 23, 162, 167 011 iron ammonia catalyst, 11-13 on manganese chromium oside, 6 011 mangnnous chromium oxide, 7 on mctds, 166. on nickel catalysts, 13-21, 165 potcntial energy curve, lit rate of, 17, 166 on tungsten, 21, 85, 165-168 van der Waals, 4, 13 on zinc oxides, 4, 6-7 aldehyde formation, 100 atomic-molecular interaction, 180 atoms, rate of production, 172-173 reaction with mcthylene groups, 98 reactivity of, 181 recombination efficiency, 173-174 role in exchange renction, 93 trapping by molybdenum oxide, 172173 on tungsten, 172-173 biological activation, 191-193 catalytic activation, 157-195 cat,alytic reaction with acetylene, 95 clieniisorption of, on ammonia catalyst, 4 on chromium oxide, 22-23 heat of, 13, 19,23, 187 on iron 11-13 on metah 158-184 molecular structure 167 nat~ureand packing 187 on iiickel 13-19,187
310
BUBJECT INDEX
Hydrogen (codinued) cracking inhibitor 225-229, 240, 243, 247, 251-255 desorption from oxidm and metals 7, 19-21,24
desorption-readsorption on chromia catalysts, 18, 23 disproportionation, 42 dissociation, 172-173, 181 effect in isomerization of pentanes, 226-229 electron diffraction, 164 exchange reaction, with chlorine, 44 with deuterium on chromium oxide 175
with deuterium, enzymatic, 193 with deuterium, in isomerization, 219, 222
with deuterium on metals, 23-24, 175, 177-179
with deuterium and methylethylenes, 94,185
with halogen, 34, 39-40, 63 with tritium, 39, 216, 217 in hydrocarbon formation, 101 ionization potential, 166 isobars of, 12-19 migration in hydrocarbon synthesis, 98 monolayer of, on tungsten, 161-163 nickel surface covering, 15 overvoltage, 174 poisoning action of, 180 readsorption on nickel-chromia catalyst, 18
readsorption on oxidea, 7 transfer, 31, 41-44, 61, 57-59, 62, 192 treatment for removal of waxy hydro-
carbons, 100 tungsten surface covering by, 21, 167 van der Wads adsorption of, 4, 13, 19 Hydrogen aluminum tetrachloride, 39, 217 Hydrogen bromide, 203404,245,253 Hydrogen bromide-aluminum bromide catalysts, 28, 204-205, 219, 245, 253 Hydrogen-carbon monoxide ratio, 130-140 Hydrogen chloride, in aluminum chloride eludges, 48 concentration effect in butane isomerization, 207
cracking inhibitor, 206-207, 225-226, 247, 255
effect in butane isomerization 206-207 promoter, 27, 31, 38, 44, 48-51, 54, 204,225 . Hydrogen chloridealuminum chloride catalysts, 28, 31, 203-206, 217, 223, 224, 229, 241-243 Hydrogen fluoride, alkylates, 82 catalyst, 28, 32, 34, 37-38, 40,43, 59-62 promoter, 31, 64 sludge, 43, 60 Hydrogen-oxygen reaction, 1%191 Hydrogen sulfide, 110 Hydrogen- catalyst, 192 Hydrogenlyase catdyst, 193 Hydrogenation, of acetylene, 96-97 of carbon disulfide, 109 of carbon monoxide, 97-100, 115-116, .see ale0 Fischer-Tropsch synthesis &-addition, 186 destructive, 106, 255 of ethylene and olefins, 92-96, 180, 186-187
catalyst for, 92, 185 intermediates, 96 mechanism, 92-95 on metal films, 187 on nickel films, 94-95 transition metals, 187 ethylenic and acetylenic bonds, 186 of fur=, 107-108 hydrocarbon synthesis from carbon monoxide, 97-100 intermolecular, 44 of methyl oleate with deuterium, 186 of pyrrole, 108 steric factor in, 95, 108 of thiophene, 108 unsaturatsd organic compounds, 95 Hydroxyaluminum dibromide catalyst, 221-222
Hydroxyaluminum dichleride, 241 I
I. G. Farbenhdustrie, ~atalyste,123-124, 127-130 hot-ges recycle procm, 132-133
SUBJECT INDEX
311
liquid-phase process, 132-133 hydrogen overvoltage on, 174 oil-reoycle process, 132-133 isobar for hydrogen on, 11-12 Industrial catalysts, 193 metauic, 280 Inhibitom, recombination efficiency of hydrogen cracking,208-207, 225-238, 240-243, atoms on, 174 247, 249, 254-255 reduction of, for adsorption messurei0 dehydrogenation, 194 men*, 11-13 of enzymatic action, 194 surface area, 85 of thiophene breakdown, 110 Iron-ammonia catalyst, Intensity of x-ray diffraction lines, 262-263 adsorption-deeorption on, 21, 11-13 Interaction energy, 168-170 desorption-readsorption on, 11-12 Intermediate hypothesis, carbide, 152 metallic iron in, 84 Intermediate pressure synthesis, 145 in oil recycle process, 124 Intermediates, promoters, 82-83 in Fischer-Tropsch synthesis, 97-98, surface measurement, 67 138-140, 149, 280 synthetic, 130, of hydrogenation, 96 Iron-carbonyl catalyst, 124 in iso-synthesis, 101 Iron-cobalt carbonyl catalyst, 136 methyl cyclohexene, 190 Iron-copper catalyst, 138-140, 160 in bluene production, 190 Iron oxide, 85,llQ Intermolecular hydrogenation, 44 Iron oxide process, 119 Internally cooled converter process, 132- Isoamyl alcohol, 188 133 Isobars of adsorption, 4-25 Interplanar spacing of solids in x-ray Isobutane, diffraction, 258-259 alkylation, 30-62, see also Alkylation of Interstitial solutions, 272 butanea Intramolecular rearrangements, 35 in n-butane isomerbation, 202-215, Ionization potential of hydrogen, 165 see alao Isomerisation of butanes Iron, catalyst, 44, 229, 240 activity of, 85 chloroalkylation, 46 carbide, 138-140, 149, 280 equilibrium constant, 214-215 catalysts, thermal alkylation, 29-30 adsorption isobsre on, 11-13 Isobutyl alcohol, 101,188 carbon deposition on, 138-140 Isobu tylene, chemiaorption of carbon monoxide on, akylation of isobutane with, 30, 54, 67 66 60 conversion to carbide, 139 alkylation of isopentane with, 42 for Fischer-Tropsch synthesis, 97-100, Isomer formation, 98 116-117, 187, 279-280 Isomerization, life of, 149 of alkanea, 201-266 Lurgi, 125 of butanes, 201-224 preparation, 129-130 catalysts, pressure coefficient of synthesis on, aluminum bromide, 202-203, 207149 211, 215, 219-221 sintered, 123 aluminum bromide-deuterium brofor synol process, 133-134 mide, 218-219 h t s , 127-129, 131 aluminum bromidehydrogen chemkorption on, 11-13 bromide, 28,204-205, 219 film of, 2 aluminum chloride, 201-203, 207, heterogeneity of surface, 11-13 209, 212, 215, 219,224
312
SUBJECT INDEX
Isonierizntion (continued) aluminum chloride-bauxite, 224 aluminum chloride-charcoal, 207,
209 aluminum chloride-hydrogen chloride, 203-206, 217, 223 hydroxyaluminum dibromide, 221222
chain mechanism, 217 commercial methods for, 223-224 exchange rcaction, 222-223 Hedman mechanism, 216-217 high vacuum technique, 203-213 isobutane in, 205-215 o l e h effect on, 203-207, 217-219 oxygcn effect on, 207-209, 219-221 Powell and Rcid mechanism, 216-21 7 products, 203, 207-213, 218-224 promotera,203-213, 217-219, 221-223 temperature effect, 203, 206 water effect, 204-213, 221-223 of heptanes, 202, 252-255 catalysts, aluminum bromide, 252-253 aluminum chloride, 252 cracking inhibitors, 254-255 destructive alkylation, 255 products, 252, 254 promoters, 253 temperature effect, 253 of hexanes, 202, 244-252 catalysts, aluminum bromide, 244-245, 249 aluminum bromide-hydrogen bromide, 245 aluminum chloride, 244 cracking suppressora, 247 equilibrium constants, 248-252 products, 244-247, 250-251 promoters, 244-245 velocity, 251-252 of pentanca, 202,224-244 autodestructive skylation, 228 CStdYSte, aluminum bromide, 225-227 aluminum chloride-hydrogen chloride, 224-225, 228-238, 241-243
hydroxyaluminum dichloride, 241 life, 231, 236
carboniuni ion, 240-242 chain mechanism, 240-242 commercial methods for, 244 cracking suppressors, 225-238, 240 Evering, d’ouville, Lien, and Waugh mechanism, 242-243 hydrogen cffect, 226-229 organic additives, effect, 229-238 paraffins, 238 products, 225-244 promoters, 224226, 228 reaction time, 226 temperature effect, 228-229 of propane, 223 reactions, 112 ring closure, 190 IsopsratEn alkylation, 27-59, see aLvo Alkylation of isoparaffins Isoparaffin synthesis, 136-138 Isopentane, 30, 225-231, 238-240, 245. 248-249
Isopropyl chloride, 44-45, 62 Isopropyl fluoride, SO, 62 IsMlyntheSis, 101, 137-138 Isotherm, adsorption, 66-89 Isotopes, 24, 85
K Kaiser Wilhelm Institute catalyst, 127-12‘3 Kelvin equation, 86 Kieselguhr catalyata, me under Cobalb kieselguhr catalysts and under Nickel-kieselguhr Catalysts Klocknerwerke plant, 121 Kogaain, 116-117 Krupp Fischer-Tropsch plant, 119 Krupp-Lurgi proccss, 119 Krypton, 78
L Langmuir-Hinahelwood concept, 158 Lattice, nickel, 93, 95 spinel, 104 Lattice constants, in x-ray diffraction 272-273
Lattice dimensions, of metal catalysts, 92 Lattice distance, in nickel-kieselguhr catalyst, 95 Lattice layer, molybdenum sulfide, 106
SUBJECT INDEX
Lattice structure, higher oxides of cobalt, nickel, manganese, 110 molybdenum, 102 molybdenum dioxide, 101 nickel sulfide, 108
Lead, 174 Liquefaction, heat of, 71 Liquid-phase process, 124 Lower layer in isoparaffin alkylation, 31, 43, see also Sludge Luetzkendorf plant, 119 Lurgi catalysts, 123-130 Lurgi recycle operation, 125-126 Lurgi recycle proccss, 132-133 Luxmaese-copper ammonia nitrate catalyst, 125
M Macromolecule formation, 98 Magnesium oxide, 116 Magnetite, 280 MaIonic acid, 194 Manganesechromium oxide, 3, 6 Manganeee dioxide, 111 Manganese oxides, higher, 110 Mangauites, 111 Manganouschromium oxide, 7 Maxima of adsorption isobars, 4, 11-13 Mettsurement of small angle of scatter, 264-272
Medium-pressure process, 118 Mercury, 174 Metal alloy catalysts, 189 Metal atoms, displacement of, 94 Metals, activated h r p t i o n on, 166 chemisorption of hydrogen on, 158-184 electric contact potential, 159 electron gas of, 171 evaporation, 165 heat of adsorption of gases on, 159 heterogeneity of surfaces, 9-22 recombination of hydrogen atoms on, 173-174
thermionic emiasion, 163 transition, 187 Methane, structure, 167 Methanedeuterium exchange, 181 Methane formation, 98, 188
3 13
Methanol synthesis, 115 Methyl bromide, 53 Methyl cyclohexene, 190 Methyl oleate, 186-187 Methyl side groups, 98 2-Methylbutane, 138, 238, 254 ZbIethyl-2-butene, 58 Methylcyclopentane, 249 Methylene blue, 191 hlethylene groups, 97-98, 188 Methylene trnnsfcr, 40 Methylethylenes, 94, 185 2-Methylhexane, 117, 253-254 3-MethyIhexanel 117, 253-254 3-Mcthiyloctane, 116 ZMethylpeutane, 30, 43, 49, 54, 117-118, 138, 244-251, 254
%Methylpentme, 117-118, 138, 244-251, 254
2-Methylpropane, 101 Middle-pressure synthesis, 121-122 Molecular-atomic hydrogen interaction, 180
Molccular-cross section of water, 78 Molecular orbital method, 105 Molcculnr structurc of chernisorbed hydrogen, 167 Molecules, adsorbed, 21, 70, 77-78, 94 in B. E. T.theory, 70 cross section areas of, 77-78 diatomic, 91-112 geometric arrangement, 21 isotopic, 24 macro-, 98 packing, 78 spncing, 94 Big-sag form, 99 Molybdates, 101 Molybdenum, catalysts, 108 lattice structure, 102 Molybdenum-cnrbon bonds, 106 Molybdenum dioxide catalysts, 101-102, 105-108
hlolybdcnum disulfidc catalysts, 106-108, 112
Molybdenum oxide catalysts, 101-102, 108, 172-173, 190
hlolybdenum sulfidc catalyst, 105-106,108
314
SUBJECT INDEX
Molybdenum trioxide, 101, 116, 259, 280-288
Monochromatars, 260-262 Monolayer, of &orbed gaa, 67-89 on adsorption isotherms, 67-68 concept, 68-71 of hydrogen on tungsten, 161-163 and multilayer theory, 68-71 reaction in, 158 Multilayer theory, 68-71 Multiplet theory, 91 Multi-atage operation of converters, 121
a Naptha, 248 Naphthenea, cracking inhibitor in isomerisation, 233-234.237-238, 249 in h y n t h e s b , 101 Neon, 160 Nickel, active centers on, 20 atoms of, 95 catalysts, acetylene polymerization on, 97 adsorption of ethylene on, 94 adsorption of carbon dioxide on, 66 adsorption of hydrogen on, 19 atom exchange on, 181 chemisorption of hydrogen on, 13-19 deuteriumethylene exchange on, 92, 184-185 deuterium-methane exchange on, 181 double-bond migration, 185 for Fbcher-Tropsch process, 97-100 heterogeneity of surface, 13-19 hydrogen isobars on, 13-21 hydrogen-oxygen reaction on, 181 in hydrogenation of ethylene, 92, 185 in hydrogenation of furan, 108 in hydrogenation of olefine, 94 in hydrogenation of pyrrol, 108 in parahydrogen conversion, 179-180 110-plane of, 97 preparation, 19 promoter, 17 rate of adsorption of hydrogen on, 13-19
spacing of atom, 84 van der Waala adsorption on, 13-21 as component of copper catalyst, 9 deeorption of hydrogen on, 14-20 diffusion of oxygen into, 166 filmof, 2, 96, 159, 166, 187, 259 heat of adeorption of hydrogen on, 1%20 heterogeneity of surface of, 19-20 in hydrogendeuterium exchange, 175, 179 lattice, 93,96 Nickel-carbon bond in ethylene adsorption, 93-95 Nickel-carbon distance, 95-96 Nickel-carbonyl, 93 Nickel-chromia catalyst, 18 Nickel hydrosilicate, 95 Nickel hydroxide, 96 Nickel-kieselguhr catalysts, adeorption on, 16-17 dispersion of atoms in, 95 lattice distancee, 95 preparation, 96 sintering of, 16 surface area, 13, 16 surface covering on, 15 Nickel-magnesia-alumin&kieselgu hr catalyst, 117 Nickel-nickel oxide, 105 Nickel nitrate, 19 Nickel oxidea, 110 Nickel subsulfide catalyst, 109-110 Nickel sulfide catalyst, 108 Nickel-thoria catalyst, 17 Nitrogen, activation of, 25 adsorption of, on carbon, 73 on charcoal, 80 isotherms, 68,76 low temperature, 67-89 on nickel-kieselguhr, 16 on poroue glass, 87-88 in surface meaauremente, 2 on tungsten, 21, 85, 116 on zinc oxide, 6 chemisorption of, 22, 167 desorption-readsorption of, 21 dilution with, la142
BUBJECT INDEX
molecular c r m section area of, 78 in parahydrogen converaion, 179 reaction with tungsten, 165 as surface covering, 15 van der Waals adsorption of, 22, 86-87 Nitroethane catalyst, 51 Nitromethane catalyst, 30, 51-62 Nitropropanealumhum chloride c a t alysts, 61-52 n-Nonyl alcohol, 136 0
315
Oxidation catalysts, 101, 111 Oxide, filmsof, 112 layers, 191 manganese, higher, 110 surfaces, 3-9, 22-26 0 x 0 synthesis, 135-136, 153 aldehyde formation, 100,136 catalysts, cobalt, 100, 135, 136 copper chromite, 130 sulfide, 136 cyclohexene, 100 hexahydrobenzaldehyde, 100 producta, 136 oxygen, adsorption isotherm, 68 adsorption on tungsten, 166, 176-177 on charcoal, 174 chemisorbed, 167 diffusion into met&, 166 effect in butane isomerization, 207-209, 219-220 heat of adsorption of, on tungetan, 162 importance for aromatization catalysts, 104 poison in parahydrogen conversion, 170 &a promoter, 174, 208 surface covering by, 177 Oxygen-hydrogen reaction, 190-191
n-Octane cyclization, 190 Octyl ions, 35 Oil recycle procese, 124 Oleh, . aldehyde formation, 100 alkylating agenta, 29 in alkylation of isoparaflhs, 42-44, 55-03 bond angles, 112 catalyst in hydrogenation of, 94 cyclic, loo exchange reactions of, 93 in Fischer-Tropach synthesis, 98 gaseow 29 higher, 93-94 hydrogenation mechanism, 92-95 in iaomerization of butanes, 203-207, 217-219 molybdenum dlulfide catalyst, 112 P polymerization of, 39,42 Palladium, 2, 179 aa promoters, 202-206,217-219 Palladium black catalyst, 191 reversible adsorption of, 39 Palladium-carbon distance for adsorbed spacing on molybdenum dioxide, 102 acetylene, 96 steric effects, 94 Palladium catalyst, Ole&-catalyst complex, 33 adsorption of ethylene on, 94 Organic additives in pentme isomerieahydrogen overvoltage on, I74 tion, 229-238 in hydrogenation of furan, 108 Organic compounds, unsaturated, 95 in hydrogenation of olefins, 94 Orientation of crystallitea in catalyate, in hydrogenation of pyrrole, 108 259 Orthohydrogen-parahydrogen conversion, in parahydrogen conversion, 180 recombination of hydrogen atoms on, 174-181, see also Parahydrogen con174 version spacing of atoms, 94 Osmium catalyst, 108 Osmium oxide supported on asbestos, 116 Palmitic acid, 80-81 Oxidation, paraffins, by acetylene polymerization, 97 biological, 191-192 cyclization, 189-190 of carbon monoxide, 110-111, 273479
316
SUBJECT INDEX
Purdins (continued) in Fischer-Tropsch synthmis, 98 in pcntanc immerizcttion, 238 Parafin waxes, synthcsis of, 115 Parahydrogen, 2 Pnrnhydrogen conversion, absolute rate of, 179-181 atomic-molecular hydrogen int erart ion, 180
calculation of, 179-181 catalysb, a,a’-diphenyl- 8-picryl hydrazyl, 176 ammonia, 182-183 Fischer-Tropsch, 187-188 nickel, 179-180 palladium, 180 paramagnetic oxides, 174-178 platinum black, 174-176 tungsten, 180 on charcoal, 174-181 fhcmicnl mechanism, 176 determining factors, 182 gaps in adsorbed layer in, 180 hydridea, 182 hydrogen-deuterium exchange, 179 011 mctale, 177, 179 paramagnetic mechanism, 154-17ti phthalocyanine in, 175 poisoning, 179-180 promoter, 176 on pyrex glase, 180 retardation by ethylene, 94 surface covering by oxygen, 177 temperature independent factor, 1’79180
tungsten, 176, 180 l’armmgnetic mechasism, 174-1 76 Parnmngnetic oxides catalyst, 174-176 Particle repulsion, 168-170 Particle size, nctivated alumina, 280-288 alumina gel, 285-288 from small angle scattering, 263-261 temperature effect on, 280-288 x-ray dctermination of, 80,263-267 l’nrticles in catalysts, x-ray studies, 280-288
Pentanes, isomerization of, see Isomerisation of pentanes 2-Pe1itene, 42, 58
Phosphopyridine nucleotides, 192 Photoelectric work function, 163 Photographic measurement of small angle scatter, 267 Phthalocyanine, 175 Physical adsorption, see van der Waals adsorption Piiitsch-Hillebrand process, 119 Plane crystal monochromators, 260-261 Platinum, 2, 179 Platinum black catalyst, 174-176, 181 Pltltinum catalyst, adsorption of ethylene on, 94 hydrogen overvoltage on, 174 hydrogen-oxygen reaction on, 191 in hydrogenation of furan, 108 in hydrogenation of o l e h , 94 in parahydrogen conversion, 180 recombination of hydrogen atoms on 174
Npacing of atoms, 94 “Point B” concept, 67-68 Poisoning, of catalyste, 2, 91, 111, 113, 194 of catalytic action, 2, 91 in parahydrogen convewion, 179-180 Polany-Eyring method, 171 Polyhydric alcohole, synthesis, 115 Polymerization, of acetylene, 96-97 cyclopentadiene in coal gas, 110 of olefins, 39, 42 Pore size, 86-88 Potaah-alumina promoter, 12 Potassium oxide promoter, 82-83 Potential barriers, 195 Potential energy curve, 171 Powder method of x-ray ditfractioii, 258-263
Powell and Reid mechanism, 216 Prcssure coefficient, 140-149 Pressure effect, on catalysts, 140, 143, 149 on contact time, 143 in Fischer-Tropsch synthesis, 140-143 in isobutane alkylation, 54 on products of iso-synthesis, 137-138 on reaction rate, 140-142 on yield, 142-144, 148 Pressure range in catalytic alkylation, 30
317
SUBJECT INDEX
Products of carbon monoxide hydrogenation,' 115-116 Promoters, 113 in alkylation, 27-28, 30-31, 38, 41-44, 48-49,54 alumina, 82-83, 111 of aluminum chloride sludges, 48 of ammonia catalyst, 12,82-83 boron fluoride, 203-204 of boron fluoride catalyst, 54 bromine, 209 bromoparaffins, 203-204 of catalytic action, 2 of cobalt catalyst, 116 copper oxide, 110 ethyl chloride, 48 of ferric oxide catalyst, 11 1 of Hopcalite catalyst, 110 hydrogen bromide, 203-204, 245, 253 hydrogen chloride, 27, 31,38, 14,48-49, 51,54, 204,225 hydrogen fluoride, 31, 54 hydrogen halide, 28, 43 in komerization of alkaues, 203-213, 217,219,223-226, 228,253 of nickel cntalyst, 17 olefins, 202-208, 217-219 oxygen, 174,207-209, 219-221 in parahydrogen conversion, 176 potash-aluminn, 12 potassium oxide, 82-83 sodium bromide, 203-204 sodium chloride, 203-204 surface covering of, 82 surface distribution, 82-85 thoria, 17 water, 54,209-213, 221-223 zinc oxide, 176 of zirconium chloride catalyst, 54 Propane, alkylation of, 29-30 isomerization of, 223 Propene, exchange reaction with deuterium, 93 in isobutane alkylation, 30-32, 39-40, 44,46,50-52, 65-57,60-62 in isopentrrne alkylation, 61 Propyl alcohol, 188 Propylene-deuterium exchange reaction, 185
Propylene trimcrs. 58 protiin, 191 Ptoteus &an& 193 Pyridinium ring, 192 Pyrrole, 108 Pyrrolidine, 108
Q
Quarternary carbon a t o m , 118 Quartz, 77 Quinone, 191
R Radioactive carbon, 130, 152 Rate of adsorption, 4-26, 1ti6 Rate of desorption, 4-28 Rate of dissociation, 181 Rate of synthesis, 140-142 Ratio of gases in Fischcr-Tropsch proccss, 118 Reaction time, effect on dccompositiou, 226 Reactions, cracking, 112 exchange, see Exchange rcttctions hydrogen-oxygen, 190-l!) I isomerization, 112 ring closure, 101 tungsten-nitrogen, 165 between two adsorbed gas layers, 158 Reactors, atmospheric pressure, 120-121 middle-pressure, 121-122 Readsorption (see also Desorptionreadsorption) of hydrogen on nickel-chromia crrttrlyst, 18
of hydrogen on oxides, 7 Recombination of atoms, 173-174 Repulsion, adsorbed particles, 168-170 Retarders, 94,110 Rheinpreussen catalyst, 127-130 Rheinpreussen plant, 120-121 Rhodium catalyst, 108, 187 Rhodium, films of, 2 Ring closure, 101-102, 190 Roughness factor of surfaces, 1G8 Ruhrbenzin plant, 121 Ruhrbenzin products of operation, 122
318
BUBJECT INDEX
Ruhrchemie p r o c a , activation energy, 142 catalysts, 118, 123, 127-130 converter, 120 development of, 115-122 operation, 123 space-time yield, 132-133 water gaa in, 116 Ruthenium catalyst, 143, 144, 148-148
s Schmalfeldt-Wintershall pro-, 119 Secondary carbon atom in alkylation, a9 Selective skylation, 61 Self-condensation in alkylation, 44 Shift catalyst, 119 Silica gel, adsorption of butane on, 72 adsorption of gases on, 69-71 catalysts, 259, 288 van der Waab adsorption on, 69-71 x-ray studies, 288 Silver catalysts, 174, 189-190 Sintering, activation energy of chemieorption, 17 influence of, on surface heterogeneity, 18-17 of niokel-kieselguhr, 16 surface shrinkage, 88 Sites on tungsten,178 Sludge, 31, 43, 40-50, 65, 60 Sodium bromide, 203-204 Sodium chloride, 203-204 Sodium chlomaluminate cataIyst, 52 Solid catalysts, x-ray diffraction in, 267-288 Solids, fine pore, 80 heterogeneity of surface, 1-26 large pore, 80 non-porous, 79-80 outgassing of, 81 palmitic acid adsorption on, 80-81 particle Sire, 80 pore size, 87-88 spacing of, 258-259 solutions in, 272 stearic acid adsorption on, 80 surface area maeaurement, 2, 3, 8 1 4 x-ray scattering on, 259, 204
Solubility of gea in metal, 168 Solutions, 272 Space-time yield, of hydrocarbons, 147 of various Fischer-Tropsch processes, 132-133
Spacing, of acetylene on metals, 96 c a t a l y h in dehydrogenation, 189 in ethylene adsorption, 185 of ethylene on nickel, 93-95 interplanar, of solids,25&259 of molecules, 94 of molybdenum oxide in cyclizstion, 190 in nickel lattice, 95 Ni-Ni dietance in nickel sulfide, 108 of olefine on molybdenum dioxide, 102 of eolids in x-ray studiee, 258-250 epinel type latticea, 104 of &aces, 194 specific activity of transition metals, 187 Specimen preparation for x-ray studies, 280
Spectrometer, x-ray, 268 Stabilizers, 103 Stearic acid, 80 Steric effects for high olefin coiicentratioiis, 94
Steric factor in hydrogenation, 05, 108 Substitutional solid solutions, 272 Subtractive solutions, 272 Succinic acid, 194 Sulfide catalysts, 136 Sulfide fiima, 112 Sulfided catalyst, 105
sulfur, 106 Sulfuric acid catalyst, in isobutane alkylation, 38, 4142, 66-69
in isomerhation, 202 in koparaffin alkylation, 28, 32-34, 38, 40-43,5560
in isopentane alkylation, 40, 57, 59 Sulfuric acid sludges, 43, 55 Surface adsorption, 171 Surface area, of barium d a t e , 77 of carbon black, 79 of catalysts, alumina, 281-285
319
SUBJECT INDElx
copper, 9 Fischer-Tropsch, 85 solid, 81-82 catalytic activity, 82 of chromium trioxide on alumina, 104 decrease by sintering, 16, 88 of different types on one adsorbent, 7-28
of glass sphcres, 80 measurement, 65-89 adsorption isotherms, 75 of ammonia catalyst, 67 by Brunauer-Emmett method, 3,4,25 by butanc adsorption, 2 butane on silica gel, 72 of charcoal, 80 of chemisorption, 66, 85 by Gregg's method, 77 by Harkins and Jura method, 75-77 limitations in, 77-78 low limit in, 81 of nickel-kieselguhr catalyst, 13, 16 outgassing of solids, for, 81 of solids, 2,3 of titanium dioxide, 76 van der Waals adsorption in, 75-77 of nonporous solids, 79-80 pore aize of a solid, 87-88 porous solids, 80 ' of quartz, 77 roughness factor, 168 of titanium dioxide, 77 of tungsten wire, 159 of zinc oxide pigments, 80 of zirconium silicata, 77 Burface carbide in Fischer Tropsch synthesis, 150 Surface covering, in ~ s t d y t i reactions, c 25 hydrogen on tungshn, 21, 167 of nickel by hydrogen and nitrogen, 15 promoters, 82 Surface deuteride on tungsten, 178 3urface hydride on tungsten, 178 3urface regions of different adsorptivo character, 5-26 g r f a c a shrinkage by sintering, 88 Surface spacing, 194 3urface spreading force of adsorbed films, 76
Sp01 catalyst, 133-134 Synol process, 133-136 Synthesis, hydrocarbons, 25, 97-101, 115-153, see also Fischer-Tropsch syntheais Synthesis, i s o p a r h , 136-138 Synthasis of methanol, 115 S y n t h d of par& waxes, 115 B y n t h b of polyhydric alcohols, 115 Synthetic gas, 119, 132-133, 150 Synthetic iron-ammonia catalyst, 130
T Techniques of x-ray studies on catalysts, 260-272
Temperature coefficient in FischerTropsch synthesis, 140-149 Temperature coefficient, hydrogenation of acetylene, 96 Temperature effect, on activated alumina, 280-288 on alumina gel, 285-288 on barium carbonate-ferric oxidc catalyst, 273-277 on butane isomerisation, 203, 206 on isobutane alkylation, 44-45, 50-51, 54-56, 60-61
on particle size, 280-288 in pentane isomerisation, 228-229 on pressure coefficient, 149 on products of iso-synthesis, 137-138 on throughput in Fischer-Tropsch synthesis, 144-145 on yield in Fischer-Tropsch syntheaie, 148
Temperature independent factor in parahydrogen conversion, 179-180 Temperature range in catalytic alkylation, 30 Tertiary aarbon atom, 28, 32, 37, 57-58, 118
Test method for catalytic activity, 281-285 Tetradeutero-ethylene, 93 Tetrahydrofuran, 108 Tetrahydrothiophene, 108 Theory, B. E. T., 70 carbide, 150
320
RUBJECT I N D E X
Theory (cantinued) of condensation of adsorbed R ~ R C Ri n capillaries, 74 of condensation in pores, 86-88 Dcbye Schemer, 261-262 multilayer, 68-71 multiplet, 91 of particle sizes, 264 for rate of production of hydrogen atoms, 172-173 Whitmore's, 35 of x-ray scattering on solids, 259. 264 Thermal alkylation, 29-30 Thermionic emission, 163 Thiophene, 108, 110 Thoria, 17 Thoria catalyst, 101, 137 Thoria dioxide, 116 Thoria-zinc oxide-alumina catalyst,, 137 Thorium nitrate, 9 Throughput' effect on yield, 145-146 Throughput, temperature effect on, 134145 Titanium dioxide, 76-77 Toluene, 189-190, 237 Transition metals, 187 Transition of molybdenum oxide, 106 Transition state, energetics, 194 Tribromopropane catalyst, 30 TrichloroacetaIdehyde catalyst, 30 1,1,2-Trichloro-3,3-dimethylbutane,46 2,2,3-Trimethylbutane, 253 Trimcthylethylene, 94 Trirnethylethylethylene,40 2,2,3-Trirnethylpentane,35-36, 44 2,2,4Trimethylpcntane, 42, 44, 49, 57, 61, 62 2,3,3-Trimcthylpentane, 57, 62 2,3,4Trimethylpcntane, 44, 57 Tritium, 216-217 Tritium butyl, 216 Tritium hydride, 216 Tritium-hydrogen exchange reaction, 39, 216-217 Tritium sulfuric acid catalyst, 38-39 Two-crystal method for measuring angle scattering, 269-271 Two-point adsorption, 91-112, 1 0 Tungsten, adsorbent and catalyst, accommodation coefficient of, 159-163
activation energies, 178-179 active and inactive planes of, 178-1 70 adsorption of cesium on, 171 of hydrogen on, 21, 85, 162, 165-168 of nitrogen on, 21-22, 85, 166 of oxygen on, 162, 166, 176-177 chemiaorp tion of gases on, I6 1 of nitrogen on, 22 chemisorption studies, 159 clean surfaces, 165 contact potentials, 16, 176-177 desorption-readsorption of nitrogen on, 21 diffusion of oxygen into, 166 in ethylene hydrogenation, 180 films of, 159, 165, 178 heterogeneity of surface, 21, 168 hydrogen atoms on, 172-173 hydrogen overvoltage on, 174 in parahydrogen conversion, 176, 180 reaction with nitrogen, 165 recombination efficiency of hydrogen atoms on, 174 sites of, 178 surface covering by hydrogcn, 2 I , If17 Tungsten deuteride, 178 Tungsten hydridc, 178 Tung&n powder, 21, 166-168 Tungsten trioxide, 116 Tungsten wires, 159
V Valency electrons, 195 Vanadium oxide catalyst, 101 van der Waals adsorption, 2, 66-89 activation energy of, 16 chemisorption, influence on, 81 differentiated from chemisorption,
2-3 of hydrogen on nickel, 19 of hydrogen on zinc oxide, 4 isotherms, 68-71 on nickel catalysts, 13-21 of nitrogen on tungsten, 22 of nitrogen on non-porous solids, 86-87 in surface mensurements, 75-77 temperature regions of, 3 transition to chernisorption, 15
321
SUBJECT INDEX
van der Wash layer, interaction with chemisorbed layer, 193 Vanadomolybdate catalyst, 101 Vinyl chloride, 46
W Water, adsorption on titanium oxide, 6 effect on aluminum bromide catalyst, 209-211, 221-223, effect on hydrogen fluoride catalyst, 60 molecular cross section of, 78 promoter in komerization, 209-213, 221-223 promoter of boron fluoride catalyst, 51 Waterdeuterium gRs exchange, 181, 183, 193 Water gas in Ruhrchemie process, 118-119 Water gas shift, 115 Whitmorc's theory, 35 Winkler process, 119 Work function, photoelectric, 163
x X-ray analyais and catalytic activity, 273-288 X-ray angle of scatter, 258-273, see also Angle of Scatter X-ray diffraction, identification by, 258-259, 273-280 lattice constants, 272-273 measurement of intensity of lines by, 262-263 patterns, of catdyata, 258-260 powder method of, 258-263 in solid catalysts, 257-288 X-ray scattering, aee d o Angle of scatter Geiger counter, 266-268 Geiger-Mueller measurement of, 265-267 Guinier method, 269-271 photographic measurement of, 265-267 principIe, 257-258 silica gel, 288 small angle, 262-263 spectrometer, 268 X-ray studies, 80, 285-288 alumina gel, 285-288 cameras, 260
X-ray studies, (continued) catalysts, barium carbonate-feiric oxide, 273277 catalytic activity, 277-279 crystallites in, 259, 280-288 Fischer-Tropsch catalysts, 274-280 iron catalysts, 279-280 particles in, 280-288 patterns, 258-260 structure, 257-288 catdyt,ic activity, 281-285 crystallite size, 262-263 detection of changed lattice constants, 272-273 interplanar spncing of solids in, 258-259 monochromators, 260-270 particle size, 80, 263-267 preparation of specimens, 260 silica gel, 288 solutions in solids, 272 techniques, 260-273 X-ray test method for catalytic activity, 281-285 Xylene, 190
z Zero-point energy, 175 Zero pressure extrapolation of adsorption isotherms, 75 Zinc-chromium oxide, 3, 6-7, 23 Zinc oxalate, 4 Zinc oxide, activation energy on, 24 adsorption of hydrogen on, 4-7 adsorption of nitrogen on, 6 catalysts, 101, 137 desorption of hydrogen on, 7 desorption of hydrogen deuteride from, 24
heterogeneity of surface, 24 hydrogendeuterium exchange on, 23-24 pigments, 80 promoter in parahydrogen conversion, 176 readsorption of hydrogen on, 7 Zirconium chloride catalyst, 54-55 Zirconium chloride-hydrogen halide catalysts, 28 Zirconium silicate, 77
This Page Intentionally Left Blank